1experiment:2 name: controller_mini3 seed: 14 mode: prisme5 model: from_checkpoint6 checkpoint: experiments/runs/mini_editor_seed01_20261030005118/checkpoint7 labels:8 train: experiments/runs/labels_mini_seed01_20261030005150/labels_train.pt9 calib: experiments/runs/labels_mini_seed01_20261030005150/labels_calib.pt10 train:11 steps: 6012 lr: 0.00113 weight_decay: 0.0114 alpha: 1.015 beta: 0.516 gamma: 0.117 huber_delta: 0.118 log_every: 2019 eval_every: 3020 seq_len: 32021 lambda_grid:22 - 0.023 - 0.324 kappa_grid:25 - 0.026 sweep_batches: 227 max_budget: 428 selection_rule: max_loss_reduction_per_ops29model:30 name: mini31 vocab_size: 12832 d_model: 9633 n_layers: 234 n_heads: 435 n_kv_heads: 236 ffn_width: 25637 max_seq_len: 32038 rope_theta: 10000.039 norm_eps: 1.0e-0540 tie_embeddings: true41 init_std: 0.0242 use_workspace: true43 w: 3244 ffn_w: 9645 n_slots: 1646 n_editor_blocks: 247 n_editor_heads: 248 r_max: 849 eta_max: 0.2550 clip_c: 1.051 clip_eps: 1.0e-0652 readout_zero_init: false53 readout_init_std: 0.254 readout_scale: 0.1196771636605262855 controller_hidden: 056 ctrl_input: h57 ctrl_use_entropy: true58 lambda_cost: 0.059 kappa_risk: 0.060 dense_writes: false61