checkpoints/best_model_metadata.json

{
  "best_step": 1100,
  "best_score": 24.940000000000126,
  "score_name": "mid_train_macro_mean_reward",
  "stage": "mid_train",
  "eval_result_path": "/workspace/starVLA/results/Checkpoints/openvla_flappy_fix_latency_1/eval/mid_train/step_1100.json",
  "aggregate": {
    "stage": "mid_train",
    "step": 1100,
    "task": "flappy",
    "model_alias": "openvla",
    "fixed_episode_seeds": true,
    "eval_seed": 42,
    "total_episodes": 5,
    "mean_reward": 24.940000000000126,
    "mean_length": 219.0,
    "std_reward": 16.381282001113483,
    "std_length": 133.51404420509476,
    "task_count": 1,
    "macro_mean_reward": 24.940000000000126,
    "macro_mean_length": 219.0,
    "distributed_eval": false
  },
  "per_latency_scores": {
    "flappy/latency_1": 24.940000000000126
  }
}