{
  "schema": "roborsi.descriptive-analysis.v1",
  "input_file": "episodes.csv",
  "input_sha256": "2238da2a7a97aabb0a19f8a9a5cf6ca4c00ef99ab3ea3d73859fe89f05fd52b6",
  "scope": "25 paired task-layout observations across four frozen source revisions; descriptive only. Model-call counts describe selected completed episodes, excluding original interrupted attempts and offline reviews.",
  "arms": {
    "astra_only": {
      "n": 25,
      "successes": 10,
      "mean_score_100": 47.46666666666667,
      "astra_calls_total": 859,
      "astra_calls_median": 33,
      "native_actions_median": 458
    },
    "astra_vla": {
      "n": 25,
      "successes": 10,
      "mean_score_100": 56.13333333333333,
      "astra_calls_total": 470,
      "astra_calls_median": 10,
      "native_actions_median": 342
    },
    "vla_only": {
      "n": 25,
      "successes": 5,
      "mean_score_100": 21.6,
      "astra_calls_total": 0,
      "astra_calls_median": 0,
      "native_actions_median": 290
    }
  },
  "comparisons": [
    {
      "left": "astra_vla",
      "right": "astra_only",
      "pairs": 25,
      "success_pairs": {
        "both_success": 4,
        "left_only_success": 6,
        "right_only_success": 6,
        "both_failure": 9
      },
      "score_pairs": {
        "left_wins": 10,
        "ties": 9,
        "right_wins": 6
      },
      "mean_score_difference": 8.666666666666668
    },
    {
      "left": "astra_vla",
      "right": "vla_only",
      "pairs": 25,
      "success_pairs": {
        "both_success": 4,
        "left_only_success": 6,
        "right_only_success": 1,
        "both_failure": 14
      },
      "score_pairs": {
        "left_wins": 14,
        "ties": 10,
        "right_wins": 1
      },
      "mean_score_difference": 34.53333333333333
    },
    {
      "left": "astra_only",
      "right": "vla_only",
      "pairs": 25,
      "success_pairs": {
        "both_success": 0,
        "left_only_success": 10,
        "right_only_success": 5,
        "both_failure": 10
      },
      "score_pairs": {
        "left_wins": 13,
        "ties": 7,
        "right_wins": 5
      },
      "mean_score_difference": 25.866666666666667
    }
  ],
  "rounds": [
    {
      "round": 1,
      "source_sha256": [
        "448a1e3944e85d8b5643cbbde0a913757e36a6cdf638df65c0b2c7681a40f998"
      ],
      "arms": {
        "astra_only": {
          "n": 5,
          "successes": 2,
          "mean_score_100": 54.666666666666664,
          "astra_calls_total": 179,
          "astra_calls_median": 40,
          "native_actions_median": 425
        },
        "astra_vla": {
          "n": 5,
          "successes": 2,
          "mean_score_100": 46.666666666666664,
          "astra_calls_total": 71,
          "astra_calls_median": 10,
          "native_actions_median": 280
        },
        "vla_only": {
          "n": 5,
          "successes": 2,
          "mean_score_100": 40.0,
          "astra_calls_total": 0,
          "astra_calls_median": 0,
          "native_actions_median": 279
        }
      }
    },
    {
      "round": 2,
      "source_sha256": [
        "d34e51e30bedeae26e2b702af2a81452f8205845310f4bc7ea3ae0702e64a08a"
      ],
      "arms": {
        "astra_only": {
          "n": 5,
          "successes": 2,
          "mean_score_100": 45.0,
          "astra_calls_total": 176,
          "astra_calls_median": 42,
          "native_actions_median": 550
        },
        "astra_vla": {
          "n": 5,
          "successes": 3,
          "mean_score_100": 66.66666666666666,
          "astra_calls_total": 54,
          "astra_calls_median": 7,
          "native_actions_median": 283
        },
        "vla_only": {
          "n": 5,
          "successes": 2,
          "mean_score_100": 40.0,
          "astra_calls_total": 0,
          "astra_calls_median": 0,
          "native_actions_median": 281
        }
      }
    },
    {
      "round": 3,
      "source_sha256": [
        "d8d7441dd0b827269f938f0befee7aaf9c218194954e35e9c3908ccfc5a05ebb"
      ],
      "arms": {
        "astra_only": {
          "n": 5,
          "successes": 4,
          "mean_score_100": 80.0,
          "astra_calls_total": 138,
          "astra_calls_median": 27,
          "native_actions_median": 295
        },
        "astra_vla": {
          "n": 5,
          "successes": 2,
          "mean_score_100": 48.66666666666667,
          "astra_calls_total": 103,
          "astra_calls_median": 25,
          "native_actions_median": 342
        },
        "vla_only": {
          "n": 5,
          "successes": 0,
          "mean_score_100": 3.0,
          "astra_calls_total": 0,
          "astra_calls_median": 0,
          "native_actions_median": 550
        }
      }
    },
    {
      "round": 4,
      "source_sha256": [
        "6bf94baf2f294d729f9e3825caafed7e39ed03d1071fee2a4ae19d8a6d4ce31d"
      ],
      "arms": {
        "astra_only": {
          "n": 5,
          "successes": 1,
          "mean_score_100": 29.666666666666668,
          "astra_calls_total": 184,
          "astra_calls_median": 42,
          "native_actions_median": 550
        },
        "astra_vla": {
          "n": 5,
          "successes": 2,
          "mean_score_100": 74.33333333333334,
          "astra_calls_total": 143,
          "astra_calls_median": 37,
          "native_actions_median": 550
        },
        "vla_only": {
          "n": 5,
          "successes": 1,
          "mean_score_100": 20.0,
          "astra_calls_total": 0,
          "astra_calls_median": 0,
          "native_actions_median": 290
        }
      }
    },
    {
      "round": 5,
      "source_sha256": [
        "6bf94baf2f294d729f9e3825caafed7e39ed03d1071fee2a4ae19d8a6d4ce31d"
      ],
      "arms": {
        "astra_only": {
          "n": 5,
          "successes": 1,
          "mean_score_100": 28.000000000000004,
          "astra_calls_total": 182,
          "astra_calls_median": 33,
          "native_actions_median": 467
        },
        "astra_vla": {
          "n": 5,
          "successes": 1,
          "mean_score_100": 44.333333333333336,
          "astra_calls_total": 99,
          "astra_calls_median": 17,
          "native_actions_median": 550
        },
        "vla_only": {
          "n": 5,
          "successes": 0,
          "mean_score_100": 5.0,
          "astra_calls_total": 0,
          "astra_calls_median": 0,
          "native_actions_median": 550
        }
      }
    }
  ]
}
