{
  "bench_version": 1,
  "track": "tabular",
  "model": "mlp_tiny_shared_activation",
  "metric_direction": "lower is better",
  "n_seeds": 8,
  "baseline": {
    "best_cfg": {
      "lr": 0.01,
      "epochs": 20
    },
    "sweep": [
      {
        "cfg": {
          "lr": 0.001,
          "epochs": 20
        },
        "mean": 17.79831099510193
      },
      {
        "cfg": {
          "lr": 0.003,
          "epochs": 20
        },
        "mean": 13.153547525405884
      },
      {
        "cfg": {
          "lr": 0.01,
          "epochs": 20
        },
        "mean": 7.754182696342468
      }
    ],
    "full": {
      "mean": 7.931941092014313,
      "std": 0.8101161551333251,
      "per_seed": [
        7.2100510597229,
        8.814193725585938,
        7.362153053283691,
        7.630332946777344,
        6.520580291748047,
        8.472823143005371,
        8.71557903289795,
        8.729815483093262
      ],
      "n": 8
    },
    "union_grid": [
      {
        "lr": 0.001,
        "epochs": 20
      },
      {
        "lr": 0.003,
        "epochs": 20
      },
      {
        "lr": 0.01,
        "epochs": 20
      }
    ]
  },
  "idea": {
    "mean": 25.546241760253906,
    "std": 1.397921331795095,
    "per_seed": [
      26.91761589050293,
      25.963476181030273,
      26.16190528869629,
      22.51058578491211,
      25.61777687072754,
      26.621234893798828,
      24.128700256347656,
      26.448638916015625
    ],
    "n": 8,
    "best_cfg": {
      "lr": 0.01,
      "epochs": 20
    }
  },
  "comparison": {
    "delta_mean": 17.614300668239594,
    "idea_wins": 0,
    "n_pairs": 8,
    "per_seed_diffs": [
      19.70756483078003,
      17.149282455444336,
      18.799752235412598,
      14.880252838134766,
      19.097196578979492,
      18.148411750793457,
      15.413121223449707,
      17.718823432922363
    ],
    "p_value": 0.0081,
    "mde": 1.4419949288736407,
    "mde_rel_pct": 18.17959705128681,
    "verdict": "idea worse (significant)",
    "system_worked": false
  },
  "mechanism_signature": {
    "idea_sweep": [
      {
        "cfg": {
          "lr": 0.001,
          "epochs": 20
        },
        "mean": 136.8510456085205
      },
      {
        "cfg": {
          "lr": 0.003,
          "epochs": 20
        },
        "mean": 72.42067241668701
      },
      {
        "cfg": {
          "lr": 0.01,
          "epochs": 20
        },
        "mean": 25.3883957862854
      }
    ],
    "selected_cfg": {
      "lr": 0.01,
      "epochs": 20
    },
    "mechanism_signature": {
      "prediction": "adjacent local gain ratio is sigma=2; output changes under input scaling",
      "sigma_predicted": 2.0,
      "slope_ratios_observed_trained_model": [
        0.49992549419403076,
        0.5001490116119385,
        0.5
      ],
      "median_ratio_observed": 0.5,
      "input_scale_2_output_abs_ratio": 1.0001696348190308,
      "probe_test_mse": 26.91761589050293,
      "confirmed": false
    },
    "track_choice": "tabular: activation is an MLP scalar nonlinearity; Friedman regression supplies the matched MLP setting."
  }
}