arch:
  H_cycles: 2
  H_layers: 4
  L_cycles: 2
  L_layers: 4
  expansion: 4
  halt_exploration_prob: 0.1
  halt_max_steps: 16
  hidden_size: 512
  loss:
    loss_type: stablemax_cross_entropy
    name: losses@ACTLossHead
  name: hrm.hrm_act_v1@HierarchicalReasoningModel_ACTV1
  num_heads: 8
  pos_encodings: rope
  puzzle_emb_ndim: 512
beta1: 0.9
beta2: 0.95
checkpoint_every_eval: true
checkpoint_path: checkpoints/Sudoku-extreme-1k-aug-1000/HierarchicalReasoningModel_ACTV1 manipulative-peacock
data_path: data/sudoku-extreme-1k-aug-1000
epochs: 40000
eval_interval: 4000
eval_save_outputs: []
global_batch_size: 768
lr: 0.0001
lr_min_ratio: 1.0
lr_warmup_steps: 2000
project_name: Sudoku-extreme-1k-aug-1000
puzzle_emb_lr: 0.0001
puzzle_emb_weight_decay: 1.0
run_name: HierarchicalReasoningModel_ACTV1 manipulative-peacock
seed: 0
weight_decay: 1.0