1experiment:2 name: controller_v1rank3 seed: 14 mode: prisme5 model: from_checkpoint6 checkpoint: experiments/runs/p4_editor_v1_Ustd02_seed01_20261029235050/checkpoint7 labels:8 train: experiments/runs/labels_v1ctrl_seed01_20261030004018/labels_train.pt9 calib: experiments/runs/labels_v1ctrl_seed01_20261030004018/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 pairwise: 1.022 listwise: 1.023 rank_temp: 0.0224 lambda_grid:25 - 0.026 - 0.0127 - 0.0328 - 0.129 - 0.330 kappa_grid:31 - 0.032 - 0.533 sweep_batches: 1034 max_budget: 835 selection_rule: max_loss_reduction_per_ops36 model_overrides:37 ctrl_input: h38 ctrl_use_entropy: true39model:40 name: tiny_p441 vocab_size: 12842 d_model: 19243 n_layers: 444 n_heads: 645 n_kv_heads: 246 ffn_width: 51247 max_seq_len: 32048 rope_theta: 10000.049 norm_eps: 1.0e-0550 tie_embeddings: true51 init_std: 0.0252 use_workspace: true53 w: 6454 ffn_w: 19255 n_slots: 1656 n_editor_blocks: 257 n_editor_heads: 258 r_max: 859 eta_max: 0.2560 clip_c: 1.061 clip_eps: 1.0e-0662 readout_zero_init: false63 readout_init_std: 0.264 readout_scale: 5.25103378295898465 controller_hidden: 066 ctrl_input: h67 ctrl_use_entropy: true68 lambda_cost: 0.069 kappa_risk: 0.070 dense_writes: false71