{
  "validation_selected_step": 1100,
  "elapsed_seconds": 42.79798544710502,
  "training": {
    "positions": 3560,
    "optimal_move_rate": 0.9817415475845337,
    "illegal_moves": 0
  },
  "validation": {
    "positions": 492,
    "optimal_move_rate": 0.9004064798355103,
    "illegal_moves": 0
  },
  "test": {
    "positions": 468,
    "optimal_move_rate": 0.9166666865348816,
    "illegal_moves": 0
  },
  "trainable_parameters": 500100,
  "changed_internal_parameters": {
    "gain": 153090,
    "bias": 165455,
    "leak": 165455
  },
  "fixed_interfaces_unchanged": true,
  "one_training_seed": true,
  "split": "D4 symmetry groups of player-relative boards, SHA256 bucket; 80/10/10 approximate",
  "device": "NVIDIA GeForce RTX 3090",
  "peak_gpu_gb": 1.013288448,
  "uniform_random_optimal_rate": 0.5691086691086691,
  "untrained_test": {
    "positions": 468,
    "optimal_move_rate": 0.6282051205635071,
    "illegal_moves": 0
  },
  "games": {
    "random_fly_first": {
      "wins": 474,
      "draws": 26,
      "losses": 0
    },
    "random_fly_second": {
      "wins": 425,
      "draws": 70,
      "losses": 5
    },
    "minimax_fly_first": {
      "wins": 0,
      "draws": 500,
      "losses": 0
    },
    "minimax_fly_second": {
      "wins": 0,
      "draws": 480,
      "losses": 20
    }
  },
  "games_note": "500 games per opponent and side, deterministic learned policy, seeded random opponent/tie choices. Complete games can include training boards; held-out position accuracy is reported separately."
}