1experiment:2 name: controller_v2s23 seed: 14 mode: prisme5 model: from_checkpoint6 checkpoint: experiments/runs/p4v2_prisme_seed02_20261030001749/checkpoint7 labels:8 train: experiments/runs/labels_v2s2_seed01_20261030004543/labels_train.pt9 calib: experiments/runs/labels_v2s2_seed01_20261030004543/labels_calib.pt10 train:11 steps: 200012 lr: 0.00113 weight_decay: 0.0114 alpha: 1.015 beta: 0.516 gamma: 0.117 huber_delta: 0.118 log_every: 10019 eval_every: 25020 seq_len: 32021 lambda_grid:22 - 0.023 - 0.0124 - 0.0325 - 0.126 - 0.327 kappa_grid:28 - 0.029 - 0.530 sweep_batches: 1031 max_budget: 832 selection_rule: max_loss_reduction_per_ops33 model_overrides:34 ctrl_input: h35 ctrl_use_entropy: true36model:37 name: tiny_p4v238 vocab_size: 12839 d_model: 25640 n_layers: 641 n_heads: 842 n_kv_heads: 443 ffn_width: 76844 max_seq_len: 32045 rope_theta: 10000.046 norm_eps: 1.0e-0547 tie_embeddings: true48 init_std: 0.0249 use_workspace: true50 w: 6451 ffn_w: 19252 n_slots: 1653 n_editor_blocks: 254 n_editor_heads: 255 r_max: 856 eta_max: 0.2557 clip_c: 1.058 clip_eps: 1.0e-0659 readout_zero_init: false60 readout_init_std: 0.261 readout_scale: 38.5145187377929762 controller_hidden: 063 ctrl_input: h64 ctrl_use_entropy: true65 lambda_cost: 0.066 kappa_risk: 0.067 dense_writes: false68