1{2 "system": "p4v2_dense_s3",3 "split": "test_compose",4 "checkpoint": "experiments/runs/p4v2_dense_seed03_20261030004544/checkpoint",5 "checkpoint_sha256": "ad0aff87189cbcd49c815e59f0adc835f1b576d111a7b0d60b7332142b4382da",6 "config_sha256": "c4e269d3e05ff208da6fe05ce1c72b11c3c69353f7d124cde990da4ff785b9ba",7 "code_commit": "unknown",8 "device": "mps",9 "precision": "fp32",10 "evaluated_at": "2026-10-30T04:07:03-0400",11 "data_sha256": "06620e1cf60b7250aac71f4fb5ee74e3705f4b9aab102e22b5c5bdbed76c4ae2",12 "exact_match": {13 "budget0": {14 "accuracy": 0.046296296296296294,15 "n": 540,16 "by_family": {17 "compose": 0.04629629629629629418 },19 "edits_per_token": 0.020 },21 "forced_b2": {22 "accuracy": 0.04814814814814815,23 "n": 540,24 "by_family": {25 "compose": 0.0481481481481481526 },27 "edits_per_token": 2.028 }29 },30 "seconds": 1.531}