{
  "bench_version": 1,
  "track": "tabular",
  "model": "mlp_tiny",
  "metric_direction": "lower is better",
  "n_seeds": 8,
  "baseline": {
    "best_cfg": {
      "lr": 0.006,
      "beta1": 0.85,
      "weight_decay": 0.0
    },
    "sweep": [
      {
        "cfg": {
          "lr": 0.0015,
          "beta1": 0.85,
          "weight_decay": 0.0
        },
        "mean": 12.617201328277588
      },
      {
        "cfg": {
          "lr": 0.0015,
          "beta1": 0.95,
          "weight_decay": 0.0
        },
        "mean": 15.512621641159058
      },
      {
        "cfg": {
          "lr": 0.003,
          "beta1": 0.85,
          "weight_decay": 0.0
        },
        "mean": 8.035151362419128
      },
      {
        "cfg": {
          "lr": 0.003,
          "beta1": 0.95,
          "weight_decay": 0.0
        },
        "mean": 12.083711862564087
      },
      {
        "cfg": {
          "lr": 0.006,
          "beta1": 0.85,
          "weight_decay": 0.0
        },
        "mean": 6.865306854248047
      },
      {
        "cfg": {
          "lr": 0.006,
          "beta1": 0.95,
          "weight_decay": 0.0
        },
        "mean": 8.458353757858276
      }
    ],
    "full": {
      "mean": 7.004796802997589,
      "std": 0.3739293172768815,
      "per_seed": [
        6.71429443359375,
        7.094020843505859,
        6.866542816162109,
        6.786369323730469,
        6.342711448669434,
        7.308849334716797,
        7.395407676696777,
        7.530178546905518
      ],
      "n": 8
    }
  },
  "idea": {
    "mean": 21.87750768661499,
    "std": 1.4977844581209827,
    "per_seed": [
      22.2957763671875,
      21.944984436035156,
      22.89360809326172,
      19.71537971496582,
      19.675622940063477,
      22.763729095458984,
      21.369779586791992,
      24.361181259155273
    ],
    "n": 8
  },
  "comparison": {
    "delta_mean": 14.872710883617401,
    "idea_wins": 0,
    "n_pairs": 8,
    "per_seed_diffs": [
      15.58148193359375,
      14.850963592529297,
      16.02706527709961,
      12.929010391235352,
      13.332911491394043,
      15.454879760742188,
      13.974371910095215,
      16.831002712249756
    ],
    "p_value": 0.0081,
    "mde": 1.1379574709779237,
    "mde_rel_pct": 16.24540301427378,
    "verdict": "idea worse (significant)",
    "system_worked": false
  },
  "mechanism_signature": {
    "mechanism_signature": {
      "predicted_response_mean": 0.3883231416550379,
      "observed_response_mean": 0.3883247250989132,
      "response_mae": 1.7781029070577825e-06,
      "update_norm_cv": 0.08146856078007694,
      "n_observations": 600,
      "prediction": "observed singular response follows t/sqrt(t^2+1), t=s/sqrt(epsilon)",
      "confirmed": true
    },
    "idea_grid": [
      {
        "cfg": {
          "lr": 0.0015,
          "beta": 0.9,
          "c": 0.001
        },
        "result": {
          "mean": 105.54728984832764,
          "std": 5.39382412585861,
          "per_seed": [
            114.30227661132812,
            109.72885131835938,
            105.97409057617188,
            96.68315124511719,
            103.86304473876953,
            104.20171356201172,
            99.61021423339844,
            110.01497650146484
          ],
          "n": 8
        }
      },
      {
        "cfg": {
          "lr": 0.003,
          "beta": 0.9,
          "c": 0.001
        },
        "result": {
          "mean": 43.18799352645874,
          "std": 3.9588922932446455,
          "per_seed": [
            49.57056427001953,
            46.739295959472656,
            42.695228576660156,
            37.679901123046875,
            39.12871551513672,
            43.38615036010742,
            39.62131118774414,
            46.68278121948242
          ],
          "n": 8
        }
      },
      {
        "cfg": {
          "lr": 0.006,
          "beta": 0.9,
          "c": 0.001
        },
        "result": {
          "mean": 21.87750768661499,
          "std": 1.4977844581209827,
          "per_seed": [
            22.2957763671875,
            21.944984436035156,
            22.89360809326172,
            19.71537971496582,
            19.675622940063477,
            22.763729095458984,
            21.369779586791992,
            24.361181259155273
          ],
          "n": 8
        }
      }
    ],
    "track_rationale": "Optimizer modification belongs on Friedman#1 tabular MLP; architecture and data are shared."
  }
}