{
  "format": "push-policy-benchmark-v2",
  "world_model_config": "4000 trajectories / 10 epochs / 3 members",
  "eval_seed": 80000,
  "episodes": 100,
  "policies": [
    {
      "name": "real PPO",
      "checkpoint": "ppo-1000.json",
      "trajectory": "benchmark-4000-real-ppo.json",
      "metrics": {
        "episodes": 100,
        "mean_final_distance": 0.07112901657819748,
        "mean_min_distance": 0.03606385737657547,
        "mean_return": 133.5593719482422,
        "mean_steps": 33.7599983215332,
        "success_rate": 0.8500000238418579
      }
    },
    {
      "name": "world-model fine-tuned (4000/10)",
      "checkpoint": "imagined-finetuned-4000e10.json",
      "trajectory": "benchmark-4000-world-finetuned.json",
      "metrics": {
        "episodes": 100,
        "mean_final_distance": 0.1763755828142166,
        "mean_min_distance": 0.13785172998905182,
        "mean_return": 90.2349624633789,
        "mean_steps": 48.27000045776367,
        "success_rate": 0.5400000214576721
      }
    },
    {
      "name": "world-model cold-start (4000/10)",
      "checkpoint": "imagined-cold-start-4000e10.json",
      "trajectory": "benchmark-4000-world-cold-start.json",
      "metrics": {
        "episodes": 100,
        "mean_final_distance": 0.4642229378223419,
        "mean_min_distance": 0.45921826362609863,
        "mean_return": -17.022123336791992,
        "mean_steps": 75.0,
        "success_rate": 0.0
      }
    }
  ]
}