1experiment:2 seed: 13 model:4 name: tiny_p4v25 vocab_size: 1286 d_model: 2567 n_layers: 68 n_heads: 89 n_kv_heads: 410 ffn_width: 76811 max_seq_len: 32012 w: 6413 ffn_w: 19214 n_slots: 1615 n_editor_heads: 216 r_max: 817 readout_zero_init: false18 readout_init_std: 0.219 n_editor_blocks: 420 init:21 backbone_checkpoint: experiments/runs/p4_tiny_backbone_v2_seed01_20261029233624/checkpoint22 freeze_backbone: true23 auto_readout_scale: true24 data:25 kind: synthetic26 tokenizer: char27 splits_dir: data/splits/synthetic_v128 loss_span: answer29 eval_splits:30 - dev_iid31 - dev_template32 - dev_depth33 - dev_compose34 budgets:35 - 036 - 137 - 238 - 439 - 840 budget_probs:41 - 0.242 - 0.243 - 0.244 - 0.245 - 0.246 budget_per: position47 train:48 batch_size: 3249 seq_len: 32050 steps: 300051 lr: 0.00152 warmup: 10053 weight_decay: 0.154 grad_clip: 1.055 log_every: 5056 eval_every: 50057 eval_batches: 1258 ckpt_every: 300059 em_budgets:60 - 061 - 162 - 263 - 864 em_max_items: 100065 name: p4v2_abl_L466 mode: prisme67model:68 name: tiny_p4v269 vocab_size: 12870 d_model: 25671 n_layers: 672 n_heads: 873 n_kv_heads: 474 ffn_width: 76875 max_seq_len: 32076 rope_theta: 10000.077 norm_eps: 1.0e-0578 tie_embeddings: true79 init_std: 0.0280 use_workspace: true81 w: 6482 ffn_w: 19283 n_slots: 1684 n_editor_blocks: 485 n_editor_heads: 286 pair_scheme: adjacent87 r_max: 888 eta_max: 0.2589 clip_c: 1.090 clip_eps: 1.0e-0691 readout_zero_init: false92 readout_init_std: 0.293 readout_scale: 1.094 controller_hidden: 095 ctrl_input: z96 ctrl_use_entropy: false97 lambda_cost: 0.098 kappa_risk: 0.099 dense_writes: false100resolved_train:101 batch_size: 32102 seq_len: 320103 steps: 3000104 lr: 0.001105 warmup: 100106 weight_decay: 0.1107 grad_clip: 1.0108 log_every: 50109 eval_every: 500110 eval_batches: 12111 ckpt_every: 3000112