{
  "bench_version": 1,
  "track": "dynamics",
  "model": "rnn_small",
  "metric_direction": "lower is better",
  "n_seeds": 8,
  "baseline": {
    "best_cfg": {
      "lr": 0.006,
      "epochs": 18
    },
    "sweep": [
      {
        "cfg": {
          "lr": 0.001,
          "epochs": 18
        },
        "mean": 0.0018504196486901492
      },
      {
        "cfg": {
          "lr": 0.003,
          "epochs": 18
        },
        "mean": 0.00105226032610517
      },
      {
        "cfg": {
          "lr": 0.006,
          "epochs": 18
        },
        "mean": 0.0007705322786932811
      }
    ],
    "full": {
      "mean": 0.0007801908141118474,
      "std": 0.000154764005468824,
      "per_seed": [
        0.0007548112771473825,
        0.0008833256433717906,
        0.0005147535703144968,
        0.0009292386239394546,
        0.0005692947888746858,
        0.0008575883111916482,
        0.0007604749407619238,
        0.0009720393572933972
      ],
      "n": 8
    }
  },
  "idea": {
    "mean": 0.0015543624467682093,
    "std": 0.0004092240958003843,
    "per_seed": [
      0.0015412603970617056,
      0.0010829424718394876,
      0.0010121178347617388,
      0.0014807143015787005,
      0.001438742270693183,
      0.0017851066077128053,
      0.0016824236372485757,
      0.0024115920532494783
    ],
    "n": 8
  },
  "comparison": {
    "delta_mean": 0.0007741716326563619,
    "idea_wins": 0,
    "n_pairs": 8,
    "per_seed_diffs": [
      0.0007864491199143231,
      0.0001996168284676969,
      0.000497364264447242,
      0.000551475677639246,
      0.0008694474818184972,
      0.000927518296521157,
      0.0009219486964866519,
      0.0014395526959560812
    ],
    "p_value": 0.0081,
    "mde": 0.00030862416250668516,
    "mde_rel_pct": 39.55752322693216,
    "verdict": "idea worse (significant)",
    "system_worked": false
  },
  "mechanism_signature": {
    "prediction": "conditioning retains 100% of selected terminal-failure trajectories; replay should increase rare-event exposure",
    "predicted_valid_fraction": 1.0,
    "observed_valid_fraction": 1.0,
    "predicted_replay_fraction": 0.5,
    "observed_replay_fraction": 0.5,
    "confirmed": true
  },
  "idea_sweep": [
    {
      "cfg": {
        "lr": 0.006,
        "epochs": 18
      },
      "result": {
        "mean": 0.001735811325488612,
        "std": 0.0007333066277889873,
        "per_seed": [
          0.001395401661284268,
          0.0006777255330234766,
          0.00144183577504009,
          0.0015068890061229467,
          0.0025884252972900867,
          0.001887585036456585,
          0.0012569716200232506,
          0.003131656674668193
        ],
        "n": 8
      }
    },
    {
      "cfg": {
        "lr": 0.001,
        "epochs": 18
      },
      "result": {
        "mean": 0.0015543624467682093,
        "std": 0.0004092240958003843,
        "per_seed": [
          0.0015412603970617056,
          0.0010829424718394876,
          0.0010121178347617388,
          0.0014807143015787005,
          0.001438742270693183,
          0.0017851066077128053,
          0.0016824236372485757,
          0.0024115920532494783
        ],
        "n": 8
      }
    },
    {
      "cfg": {
        "lr": 0.003,
        "epochs": 18
      },
      "result": {
        "mean": 0.001693843849352561,
        "std": 0.0009678744347819918,
        "per_seed": [
          0.0007645400473847985,
          0.0025881300680339336,
          0.0006857462576590478,
          0.0029576336964964867,
          0.0006684368127025664,
          0.0008722841739654541,
          0.0022166145499795675,
          0.002797365188598633
        ],
        "n": 8
      }
    }
  ],
  "runtime_sec": 69.70142459869385
}