{
  "bench_version": 1,
  "track": "tabular",
  "model": "mlp_tiny",
  "metric_direction": "lower is better",
  "n_seeds": 8,
  "baseline": {
    "best_cfg": {
      "lr": 0.01,
      "rho": 0.0,
      "penalty": 0.0,
      "epochs": 20
    },
    "sweep": [
      {
        "cfg": {
          "lr": 0.001,
          "rho": 0.0,
          "penalty": 0.0,
          "epochs": 20
        },
        "mean": 215.17152404785156
      },
      {
        "cfg": {
          "lr": 0.001,
          "rho": 0.05,
          "penalty": 0.1,
          "epochs": 20
        },
        "mean": 215.17152404785156
      },
      {
        "cfg": {
          "lr": 0.001,
          "rho": 0.2,
          "penalty": 0.8,
          "epochs": 20
        },
        "mean": 215.17152404785156
      },
      {
        "cfg": {
          "lr": 0.003,
          "rho": 0.0,
          "penalty": 0.0,
          "epochs": 20
        },
        "mean": 72.50996971130371
      },
      {
        "cfg": {
          "lr": 0.003,
          "rho": 0.05,
          "penalty": 0.1,
          "epochs": 20
        },
        "mean": 72.50996971130371
      },
      {
        "cfg": {
          "lr": 0.003,
          "rho": 0.2,
          "penalty": 0.8,
          "epochs": 20
        },
        "mean": 72.50996971130371
      },
      {
        "cfg": {
          "lr": 0.01,
          "rho": 0.0,
          "penalty": 0.0,
          "epochs": 20
        },
        "mean": 24.599658966064453
      },
      {
        "cfg": {
          "lr": 0.01,
          "rho": 0.05,
          "penalty": 0.1,
          "epochs": 20
        },
        "mean": 24.599658966064453
      },
      {
        "cfg": {
          "lr": 0.01,
          "rho": 0.2,
          "penalty": 0.8,
          "epochs": 20
        },
        "mean": 24.599658966064453
      }
    ],
    "full": {
      "mean": 31.372122287750244,
      "std": 9.015775011632964,
      "per_seed": [
        22.099992752075195,
        30.866003036499023,
        23.188148498535156,
        22.244491577148438,
        28.95732879638672,
        48.877689361572266,
        33.53181076049805,
        41.21151351928711
      ],
      "n": 8
    },
    "method": "FedAvg (central mean after local minibatch step)"
  },
  "idea": {
    "mean": 42.09156918525696,
    "std": 19.275065768488197,
    "per_seed": [
      72.97396087646484,
      40.06073760986328,
      67.2032470703125,
      53.84383010864258,
      21.086515426635742,
      30.523590087890625,
      33.312156677246094,
      17.728515625
    ],
    "n": 8
  },
  "comparison": {
    "delta_mean": 10.719446897506714,
    "idea_wins": 4,
    "n_pairs": 8,
    "per_seed_diffs": [
      50.87396812438965,
      9.194734573364258,
      44.015098571777344,
      31.59933853149414,
      -7.870813369750977,
      -18.35409927368164,
      -0.21965408325195312,
      -23.48299789428711
    ],
    "p_value": 0.2999,
    "mde": 23.733515768986862,
    "mde_rel_pct": 75.65161053274996,
    "verdict": "no significant win",
    "system_worked": false
  },
  "mechanism_signature": {
    "mechanism": "trained four-worker tree residual and dual correction measurements",
    "predicted": "redistributed corrections should be nonzero and reduce stale-consensus error",
    "observed": [
      {
        "seed": 0,
        "final_residual": 7.351737976074219,
        "correction_norm": 0.4109642207622528,
        "initial_residual": 9.530355453491211
      },
      {
        "seed": 1,
        "final_residual": 7.59358024597168,
        "correction_norm": 0.3979034423828125,
        "initial_residual": 9.610031127929688
      },
      {
        "seed": 2,
        "final_residual": 7.465011119842529,
        "correction_norm": 0.3986107110977173,
        "initial_residual": 9.460624694824219
      },
      {
        "seed": 3,
        "final_residual": 7.670905590057373,
        "correction_norm": 0.40616631507873535,
        "initial_residual": 9.607115745544434
      },
      {
        "seed": 4,
        "final_residual": 7.9027018547058105,
        "correction_norm": 0.37859806418418884,
        "initial_residual": 9.442831039428711
      },
      {
        "seed": 5,
        "final_residual": 7.862105369567871,
        "correction_norm": 0.386971652507782,
        "initial_residual": 9.534515380859375
      },
      {
        "seed": 6,
        "final_residual": 7.549962043762207,
        "correction_norm": 0.40791869163513184,
        "initial_residual": 9.64505386352539
      },
      {
        "seed": 7,
        "final_residual": 7.712738037109375,
        "correction_norm": 0.4060078263282776,
        "initial_residual": 9.527835845947266
      }
    ],
    "confirmed": true
  },
  "idea_sweep": [
    {
      "cfg": {
        "lr": 0.01,
        "rho": 0.0,
        "penalty": 0.0,
        "epochs": 20
      },
      "result": {
        "mean": 47.336907148361206,
        "std": 28.887751826226218,
        "per_seed": [
          22.092214584350586,
          35.96384048461914,
          73.7575454711914,
          22.593856811523438,
          88.00531005859375,
          27.3686466217041,
          89.42167663574219,
          19.49216651916504
        ],
        "n": 8
      }
    },
    {
      "cfg": {
        "lr": 0.01,
        "rho": 0.05,
        "penalty": 0.1,
        "epochs": 20
      },
      "result": {
        "mean": 42.09156918525696,
        "std": 19.275065768488197,
        "per_seed": [
          72.97396087646484,
          40.06073760986328,
          67.2032470703125,
          53.84383010864258,
          21.086515426635742,
          30.523590087890625,
          33.312156677246094,
          17.728515625
        ],
        "n": 8
      }
    },
    {
      "cfg": {
        "lr": 0.01,
        "rho": 0.2,
        "penalty": 0.8,
        "epochs": 20
      },
      "result": {
        "mean": 43.77399754524231,
        "std": 15.561092309534422,
        "per_seed": [
          28.526273727416992,
          48.15719223022461,
          52.14424514770508,
          37.76236343383789,
          71.31439208984375,
          55.57741928100586,
          18.206344604492188,
          38.50374984741211
        ],
        "n": 8
      }
    }
  ],
  "selection_note": "Baseline and idea use the same lr/rho/penalty union; baseline ignores tree knobs as standard FedAvg."
}