1experiment:2 name: p4_overfit64_b83 seed: 14 mode: prisme5 model:6 name: tiny_p47 vocab_size: 1288 d_model: 1929 n_layers: 410 n_heads: 611 n_kv_heads: 212 ffn_width: 51213 max_seq_len: 32014 w: 6415 ffn_w: 19216 n_slots: 1617 n_editor_heads: 218 r_max: 819 init:20 backbone_checkpoint: experiments/runs/p4_tiny_backbone_seed01_20261029232626/checkpoint21 freeze_backbone: true22 auto_readout_scale: true23 data:24 kind: synthetic25 tokenizer: char26 splits_dir: data/splits/synthetic_v127 loss_span: answer28 eval_splits:29 - dev_iid30 max_train_items: 6431 budgets:32 - 833 budget_probs:34 - 1.035 budget_per: position36 train:37 batch_size: 3238 seq_len: 32039 steps: 60040 lr: 0.00141 warmup: 10042 weight_decay: 0.143 grad_clip: 1.044 log_every: 5045 eval_every: 10046 eval_batches: 447 ckpt_every: 150048 em_budgets:49 - 050 - 251 - 852 em_max_items: 40053 model_overrides:54 readout_zero_init: false55 readout_init_std: 0.556 final_exact_match: false57model:58 name: tiny_p459 vocab_size: 12860 d_model: 19261 n_layers: 462 n_heads: 663 n_kv_heads: 264 ffn_width: 51265 max_seq_len: 32066 rope_theta: 10000.067 norm_eps: 1.0e-0568 tie_embeddings: true69 init_std: 0.0270 use_workspace: true71 w: 6472 ffn_w: 19273 n_slots: 1674 n_editor_blocks: 275 n_editor_heads: 276 r_max: 877 eta_max: 0.2578 clip_c: 1.079 clip_eps: 1.0e-0680 readout_zero_init: false81 readout_init_std: 0.582 readout_scale: 1.083 controller_hidden: 084 lambda_cost: 0.085 kappa_risk: 0.086 dense_writes: false87resolved_train:88 batch_size: 3289 seq_len: 32090 steps: 60091 lr: 0.00192 warmup: 10093 weight_decay: 0.194 grad_clip: 1.095 log_every: 5096 eval_every: 10097 eval_batches: 498 ckpt_every: 150099