training/checkpoints/results_trained/summary.json

{
  "policy_label": "trained-cos",
  "use_rag": false,
  "checkpoint_load": "ok",
  "tasks": {
    "easy_brief": {
      "terminal_score": 0.5207,
      "success": true,
      "steps": 8,
      "cumulative_reward": 0.4307
    },
    "medium_brief": {
      "terminal_score": 0.6229,
      "success": true,
      "steps": 10,
      "cumulative_reward": 0.5129
    },
    "hard_brief": {
      "terminal_score": 0.6185,
      "success": true,
      "steps": 12,
      "cumulative_reward": 0.4085
    },
    "expert_brief": {
      "terminal_score": 0.683,
      "success": true,
      "steps": 14,
      "cumulative_reward": 0.443
    }
  },
  "mean_terminal": 0.6113,
  "export_dir": "/home/uchihakamal/Desktop/hackathon/round-2/autodatalab-plus/training/exports/trained_cos_4tasks"
}