Checkpoint iteration 30
Browse files- metadata_iter_030.json +75 -0
metadata_iter_030.json
ADDED
|
@@ -0,0 +1,75 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"iteration": 30,
|
| 3 |
+
"timestamp": "2026-03-01T03:01:42.653381+00:00",
|
| 4 |
+
"buffer_size": 50000,
|
| 5 |
+
"config": {
|
| 6 |
+
"iterations": 40,
|
| 7 |
+
"episodes_per_iter": 70,
|
| 8 |
+
"mcts_sims": 420,
|
| 9 |
+
"c_puct": 1.5,
|
| 10 |
+
"temp_threshold": 15,
|
| 11 |
+
"add_noise": true,
|
| 12 |
+
"seed": 42,
|
| 13 |
+
"verbose_logs": false,
|
| 14 |
+
"episode_log_every": 0,
|
| 15 |
+
"epochs": 5,
|
| 16 |
+
"batch_size": 96,
|
| 17 |
+
"learning_rate": 0.0009,
|
| 18 |
+
"weight_decay": 0.0001,
|
| 19 |
+
"buffer_size": 50000,
|
| 20 |
+
"val_split": 0.1,
|
| 21 |
+
"d_model": 128,
|
| 22 |
+
"nhead": 8,
|
| 23 |
+
"num_layers": 6,
|
| 24 |
+
"dim_feedforward": 512,
|
| 25 |
+
"dropout": 0.1,
|
| 26 |
+
"log_dir": "logs",
|
| 27 |
+
"checkpoint_dir": "checkpoints",
|
| 28 |
+
"save_every": 3,
|
| 29 |
+
"keep_last_n_local_checkpoints": 2,
|
| 30 |
+
"keep_last_n_log_versions": 1,
|
| 31 |
+
"keep_last_n_hf_checkpoints": 3,
|
| 32 |
+
"onnx_path": "ataxx_model.onnx",
|
| 33 |
+
"export_onnx": false,
|
| 34 |
+
"hf_enabled": true,
|
| 35 |
+
"hf_repo_id": "dieg0code/ataxx-zero",
|
| 36 |
+
"hf_token_env": "HF_TOKEN",
|
| 37 |
+
"hf_local_dir": "hf_checkpoints",
|
| 38 |
+
"show_progress_bar": false,
|
| 39 |
+
"trainer_log_every_n_steps": 100,
|
| 40 |
+
"num_workers": 3,
|
| 41 |
+
"persistent_workers": true,
|
| 42 |
+
"strict_probs": false,
|
| 43 |
+
"trainer_devices": 1,
|
| 44 |
+
"trainer_strategy": "auto",
|
| 45 |
+
"trainer_precision": "16-mixed",
|
| 46 |
+
"trainer_benchmark": true,
|
| 47 |
+
"mcts_use_amp": true,
|
| 48 |
+
"opponent_self_prob": 0.85,
|
| 49 |
+
"opponent_heuristic_prob": 0.12,
|
| 50 |
+
"opponent_random_prob": 0.03,
|
| 51 |
+
"opponent_heuristic_level": "normal",
|
| 52 |
+
"opponent_heuristic_easy_prob": 0.05,
|
| 53 |
+
"opponent_heuristic_normal_prob": 0.2,
|
| 54 |
+
"opponent_heuristic_hard_prob": 0.75,
|
| 55 |
+
"model_side_swap_prob": 0.5,
|
| 56 |
+
"eval_enabled": true,
|
| 57 |
+
"eval_every": 3,
|
| 58 |
+
"eval_games": 12,
|
| 59 |
+
"eval_sims": 220,
|
| 60 |
+
"eval_heuristic_level": "hard",
|
| 61 |
+
"selfplay_workers": 1,
|
| 62 |
+
"quiet_mode": true
|
| 63 |
+
},
|
| 64 |
+
"stats": {
|
| 65 |
+
"replay_size": 50000,
|
| 66 |
+
"best_eval_score": 0.0,
|
| 67 |
+
"games": 12,
|
| 68 |
+
"wins": 0,
|
| 69 |
+
"losses": 12,
|
| 70 |
+
"draws": 0,
|
| 71 |
+
"score": 0.0,
|
| 72 |
+
"heuristic_level": "hard",
|
| 73 |
+
"sims": 220
|
| 74 |
+
}
|
| 75 |
+
}
|