1experiment:2 name: p4_editor_v1_scaled3 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 - dev_template31 - dev_depth32 - dev_compose33 budgets:34 - 035 - 136 - 237 - 438 - 839 budget_probs:40 - 0.241 - 0.242 - 0.243 - 0.244 - 0.245 budget_per: position46 train:47 batch_size: 3248 seq_len: 32049 steps: 300050 lr: 0.00151 warmup: 10052 weight_decay: 0.153 grad_clip: 1.054 log_every: 5055 eval_every: 50056 eval_batches: 857 ckpt_every: 150058 em_budgets:59 - 060 - 261 - 862 em_max_items: 40063model:64 name: tiny_p465 vocab_size: 12866 d_model: 19267 n_layers: 468 n_heads: 669 n_kv_heads: 270 ffn_width: 51271 max_seq_len: 32072 rope_theta: 10000.073 norm_eps: 1.0e-0574 tie_embeddings: true75 init_std: 0.0276 use_workspace: true77 w: 6478 ffn_w: 19279 n_slots: 1680 n_editor_blocks: 281 n_editor_heads: 282 r_max: 883 eta_max: 0.2584 clip_c: 1.085 clip_eps: 1.0e-0686 readout_zero_init: true87 readout_scale: 1.088 controller_hidden: 089 lambda_cost: 0.090 kappa_risk: 0.091 dense_writes: false92resolved_train:93 batch_size: 3294 seq_len: 32095 steps: 300096 lr: 0.00197 warmup: 10098 weight_decay: 0.199 grad_clip: 1.0100 log_every: 50101 eval_every: 500102 eval_batches: 8103 ckpt_every: 1500104