{
  "bench_version": 1,
  "track": "dynamics",
  "model": "rnn_small",
  "metric_direction": "lower is better",
  "n_seeds": 8,
  "baseline": {
    "best_cfg": {
      "lr": 0.01,
      "weight_decay": 0.0001
    },
    "sweep": [
      {
        "cfg": {
          "lr": 0.001,
          "weight_decay": 0.0
        },
        "mean": 0.004298779065720737
      },
      {
        "cfg": {
          "lr": 0.001,
          "weight_decay": 0.0001
        },
        "mean": 0.0042995543335564435
      },
      {
        "cfg": {
          "lr": 0.003,
          "weight_decay": 0.0
        },
        "mean": 0.0034615940821822733
      },
      {
        "cfg": {
          "lr": 0.003,
          "weight_decay": 0.0001
        },
        "mean": 0.003432174737099558
      },
      {
        "cfg": {
          "lr": 0.01,
          "weight_decay": 0.0
        },
        "mean": 0.0014865802077110857
      },
      {
        "cfg": {
          "lr": 0.01,
          "weight_decay": 0.0001
        },
        "mean": 0.0014641905727330595
      }
    ],
    "full": {
      "mean": 0.0014106978487689048,
      "std": 0.00038184104551606325,
      "per_seed": [
        0.0009313806658610702,
        0.0018688973505049944,
        0.0010692982468754053,
        0.0019871860276907682,
        0.0009491611272096634,
        0.0014028067234903574,
        0.0016651722835376859,
        0.0014116803649812937
      ],
      "n": 8
    }
  },
  "idea": {
    "mean": 0.0014106978487689048,
    "std": 0.00038184104551606325,
    "per_seed": [
      0.0009313806658610702,
      0.0018688973505049944,
      0.0010692982468754053,
      0.0019871860276907682,
      0.0009491611272096634,
      0.0014028067234903574,
      0.0016651722835376859,
      0.0014116803649812937
    ],
    "n": 8
  },
  "comparison": {
    "delta_mean": 0.0,
    "idea_wins": 0,
    "n_pairs": 8,
    "per_seed_diffs": [
      0.0,
      0.0,
      0.0,
      0.0,
      0.0,
      0.0,
      0.0,
      0.0
    ],
    "p_value": 1.0,
    "mde": 0.0,
    "mde_rel_pct": 0.0,
    "verdict": "no measurable effect",
    "system_worked": false
  },
  "mechanism_signature": {
    "definition": "trained-model endpoint work proxy vs endpoint squared error",
    "mean_work_proxy": -0.09426483511924744,
    "std_work_proxy": 0.24112743139266968,
    "mean_test_mse": 0.0009313806076534092,
    "corr_work_error": -0.5097475039344403,
    "predicted_direction": "work penalty should reduce endpoint energy change",
    "observed_direction": "reduced",
    "confirmed": true
  },
  "idea_sweep": [
    {
      "cfg": {
        "lr": 0.01,
        "weight_decay": 0.0001,
        "lambda": 0.0
      },
      "result": {
        "mean": 0.0014106978487689048,
        "std": 0.00038184104551606325,
        "per_seed": [
          0.0009313806658610702,
          0.0018688973505049944,
          0.0010692982468754053,
          0.0019871860276907682,
          0.0009491611272096634,
          0.0014028067234903574,
          0.0016651722835376859,
          0.0014116803649812937
        ],
        "n": 8
      }
    },
    {
      "cfg": {
        "lr": 0.01,
        "weight_decay": 0.0001,
        "lambda": 0.01
      },
      "result": {
        "mean": 0.0022155136612127535,
        "std": 0.0009111505580800969,
        "per_seed": [
          0.0026169263292104006,
          0.0008814591565169394,
          0.002344637643545866,
          0.001455318764783442,
          0.0015512269455939531,
          0.003643145551905036,
          0.0034239226952195168,
          0.0018074722029268742
        ],
        "n": 8
      }
    },
    {
      "cfg": {
        "lr": 0.01,
        "weight_decay": 0.0001,
        "lambda": 0.05
      },
      "result": {
        "mean": 0.020211070426739752,
        "std": 0.0041872810874073125,
        "per_seed": [
          0.0175981055945158,
          0.014253576286137104,
          0.024302082136273384,
          0.020239518955349922,
          0.014925491064786911,
          0.025781074538826942,
          0.02489893138408661,
          0.019689783453941345
        ],
        "n": 8
      }
    },
    {
      "cfg": {
        "lr": 0.001,
        "weight_decay": 0.0001,
        "lambda": 0.05
      },
      "result": {
        "mean": 0.03044215077534318,
        "std": 0.008291443683791245,
        "per_seed": [
          0.045904748141765594,
          0.022312993183732033,
          0.031132273375988007,
          0.03332684561610222,
          0.03957861661911011,
          0.025154752656817436,
          0.025864429771900177,
          0.020262546837329865
        ],
        "n": 8
      }
    },
    {
      "cfg": {
        "lr": 0.003,
        "weight_decay": 0.0001,
        "lambda": 0.05
      },
      "result": {
        "mean": 0.031030800193548203,
        "std": 0.008264055684726617,
        "per_seed": [
          0.030995534732937813,
          0.030705397948622704,
          0.023237835615873337,
          0.049849774688482285,
          0.0321778804063797,
          0.034446511417627335,
          0.022313375025987625,
          0.024520091712474823
        ],
        "n": 8
      }
    }
  ],
  "protocol": {
    "epochs": 12,
    "n_train": 400,
    "n_test": 200,
    "batch": 128,
    "paired_seeds": [
      0,
      1,
      2,
      3,
      4,
      5,
      6,
      7
    ],
    "structural_match": "dynamics/control"
  }
}