{
  "bench_version": 1,
  "track": "distribution_token_sequence",
  "model": "custom_poolnet",
  "metric_direction": "lower is better",
  "n_seeds": 8,
  "baseline": {
    "best_cfg": {
      "lr": 0.006,
      "epochs": 12,
      "delta": 0.5
    },
    "sweep": [
      {
        "cfg": {
          "lr": 0.001,
          "epochs": 12,
          "delta": 0.5
        },
        "mean": 0.11337076500058174
      },
      {
        "cfg": {
          "lr": 0.001,
          "epochs": 12,
          "delta": 0.8
        },
        "mean": 0.11337076500058174
      },
      {
        "cfg": {
          "lr": 0.001,
          "epochs": 12,
          "delta": 1.2
        },
        "mean": 0.11337076500058174
      },
      {
        "cfg": {
          "lr": 0.003,
          "epochs": 12,
          "delta": 0.5
        },
        "mean": 0.09198673628270626
      },
      {
        "cfg": {
          "lr": 0.003,
          "epochs": 12,
          "delta": 0.8
        },
        "mean": 0.09198673628270626
      },
      {
        "cfg": {
          "lr": 0.003,
          "epochs": 12,
          "delta": 1.2
        },
        "mean": 0.09198673628270626
      },
      {
        "cfg": {
          "lr": 0.006,
          "epochs": 12,
          "delta": 0.5
        },
        "mean": 0.06806101836264133
      },
      {
        "cfg": {
          "lr": 0.006,
          "epochs": 12,
          "delta": 0.8
        },
        "mean": 0.06806101836264133
      },
      {
        "cfg": {
          "lr": 0.006,
          "epochs": 12,
          "delta": 1.2
        },
        "mean": 0.06806101836264133
      }
    ],
    "full": {
      "mean": 0.06672146823257208,
      "std": 0.011285819903738186,
      "per_seed": [
        0.06364592909812927,
        0.0627693384885788,
        0.08224158734083176,
        0.06358721852302551,
        0.05995246767997742,
        0.08138634264469147,
        0.04605041444301605,
        0.07413844764232635
      ],
      "n": 8
    }
  },
  "idea": {
    "mean": 0.018798072007484734,
    "std": 0.0037972447648455363,
    "per_seed": [
      0.017789283767342567,
      0.02162032574415207,
      0.024604864418506622,
      0.01876969449222088,
      0.015490574762225151,
      0.011365094222128391,
      0.019505014643073082,
      0.02123972401022911
    ],
    "n": 8
  },
  "comparison": {
    "delta_mean": -0.047923396225087345,
    "idea_wins": 8,
    "n_pairs": 8,
    "per_seed_diffs": [
      -0.045856645330786705,
      -0.04114901274442673,
      -0.057636722922325134,
      -0.044817524030804634,
      -0.044461892917752266,
      -0.07002124842256308,
      -0.02654539979994297,
      -0.052898723632097244
    ],
    "p_value": 0.0081,
    "mde": 0.010652952386931449,
    "mde_rel_pct": 15.96630390356262,
    "verdict": "idea better (significant)",
    "system_worked": true
  },
  "mechanism_signature": {
    "prediction": "a gross token should shift Huber pooled quantiles less than arithmetic pooling",
    "predicted_bounded_score": 0.06666666666666667,
    "observed_huber_shift": 0.0727272778749466,
    "observed_mean_shift": 0.8333333730697632,
    "confirmed": true,
    "custom_track": {
      "name": "distribution_token_sequence",
      "file": "hub_bench.py",
      "domain": "sequence"
    },
    "idea_grid": [
      {
        "lr": 0.006,
        "epochs": 12,
        "delta": 0.5,
        "mean": 0.020211950060911477
      },
      {
        "lr": 0.006,
        "epochs": 12,
        "delta": 0.8,
        "mean": 0.018798072007484734
      },
      {
        "lr": 0.006,
        "epochs": 12,
        "delta": 1.2,
        "mean": 0.020812171744182706
      }
    ],
    "selected_idea_cfg": {
      "lr": 0.006,
      "epochs": 12,
      "delta": 0.8
    }
  }
}