run:
  experiment_name: lost-cities-deep-cfr-selfplay-full-depth-slot-playability
  seed: 79
  max_iterations: null
  max_minutes: 240
  device: cuda
  use_amp: false

rules:
  n_colors: 5
  n_ranks: 9
  min_rank: 2
  n_handshakes: 3
  hand_size: 8
  expedition_penalty: -20
  bonus_threshold: 8
  bonus_amount: 20

encoding:
  derived_playability: true
  slot_aware_playability: true

network:
  hidden_size: 256
  num_layers: 3
  activation: relu

traversal:
  traversals_per_player: 70
  strategy_sample_interval: 1
  store_strategy_on_opponent_nodes: false
  store_strategy_on_traverser_nodes: true
  max_depth: null
  max_nodes_per_traversal: 1000
  opponent_policy: self_play_league
  cutoff_value_mode: score_diff
  cutoff_rollouts: 0
  cutoff_rollout_policy: random
  cutoff_rollout_max_steps: 300
  progress_every_traversals: 10
  num_workers: 8
  worker_chunk_size: 8
  regret_matching_epsilon: 0.0001
  outcome_sampling_epsilon: 0.2
  outcome_sampling_value_clip: 500
  outcome_unsampled_regret: zero
  endpoint_depth_bucket_width: 100
  endpoint_depth_bucket_max: 1000

self_play:
  current_weight: 0.5
  recent_weight: 0.3
  older_weight: 0.2
  anchor_weight: 0.0
  recent_window: 5
  max_snapshots: 20
  snapshot_every: 1

optimization:
  advantage_batch_size: 1024
  strategy_batch_size: 1024
  advantage_updates_per_iteration: 256
  strategy_updates_per_iteration: 256
  learning_rate: 0.00003
  weight_decay: 0.0001
  grad_clip: 1.0

memory:
  advantage_capacity: 2000000
  strategy_capacity: 2000000

evaluation:
  eval_every: 5
  games: 100
  batch_size: 64
  device: trainer
  num_workers: 4
  max_steps: 1000
  on_max_steps: score_diff
  opponents:
    - random
    - passive_discard
    - safe_heuristic
    - safe_heuristic_loose
    - safe_heuristic_strict
    - noisy_safe

checkpoint:
  save_every: 10
  progress_interval_seconds: 20.0