{
  "bench_version": 1,
  "track": "tabular",
  "model": "mlp_tiny",
  "metric_direction": "lower is better",
  "n_seeds": 8,
  "baseline": {
    "best_cfg": {
      "lr": 0.005,
      "weight_decay": 0.0001
    },
    "sweep": [
      {
        "cfg": {
          "lr": 0.001,
          "weight_decay": 0.0
        },
        "mean": 107.39756393432617
      },
      {
        "cfg": {
          "lr": 0.001,
          "weight_decay": 0.0001
        },
        "mean": 107.39925003051758
      },
      {
        "cfg": {
          "lr": 0.002,
          "weight_decay": 0.0
        },
        "mean": 44.976237297058105
      },
      {
        "cfg": {
          "lr": 0.002,
          "weight_decay": 0.0001
        },
        "mean": 44.978708267211914
      },
      {
        "cfg": {
          "lr": 0.003,
          "weight_decay": 0.0
        },
        "mean": 24.11869478225708
      },
      {
        "cfg": {
          "lr": 0.003,
          "weight_decay": 0.0001
        },
        "mean": 24.120020389556885
      },
      {
        "cfg": {
          "lr": 0.005,
          "weight_decay": 0.0
        },
        "mean": 21.456337451934814
      },
      {
        "cfg": {
          "lr": 0.005,
          "weight_decay": 0.0001
        },
        "mean": 21.456222534179688
      }
    ],
    "full": {
      "mean": 21.57731342315674,
      "std": 1.5687495825188427,
      "per_seed": [
        22.156478881835938,
        22.034189224243164,
        22.254423141479492,
        19.379798889160156,
        19.08241081237793,
        22.424152374267578,
        21.120243072509766,
        24.166810989379883
      ],
      "n": 8
    }
  },
  "idea": {
    "mean": 23.979673385620117,
    "std": 1.9050116210009282,
    "per_seed": [
      25.87602996826172,
      24.822738647460938,
      24.548673629760742,
      21.082088470458984,
      21.36577033996582,
      24.733022689819336,
      22.76095199584961,
      26.64811134338379
    ],
    "n": 8
  },
  "comparison": {
    "delta_mean": 2.402359962463379,
    "idea_wins": 0,
    "n_pairs": 8,
    "per_seed_diffs": [
      3.7195510864257812,
      2.7885494232177734,
      2.29425048828125,
      1.7022895812988281,
      2.2833595275878906,
      2.308870315551758,
      1.6407089233398438,
      2.4813003540039062
    ],
    "p_value": 0.0081,
    "mde": 0.5469499653262384,
    "mde_rel_pct": 2.534838117238695,
    "verdict": "idea worse (significant)",
    "system_worked": false
  },
  "mechanism_signature": {
    "track_choice": "optimizer intervention matches tabular Friedman#1 track; identical mlp_tiny systems differ only in matrix momentum state",
    "idea_config": {
      "lr": 0.003,
      "weight_decay": 0.0001,
      "beta_fast": 0.9,
      "beta_slow": 0.99,
      "weight_fast": 0.5
    },
    "mechanism_signature": {
      "prediction": "impulse response initialized by a measured trained-model gradient has decay slope between fast and slow log betas",
      "predicted_interval": [
        -0.10536051565782628,
        -0.01005033585350145
      ],
      "observed_mean_slope": -0.09078844601504563,
      "per_seed": [
        {
          "seed": 0,
          "observed_mix_log_slope": -0.09078844601504561,
          "predicted_fast_log_beta": -0.10536051565782628,
          "predicted_slow_log_beta": -0.01005033585350145
        },
        {
          "seed": 1,
          "observed_mix_log_slope": -0.09078844601504557,
          "predicted_fast_log_beta": -0.10536051565782628,
          "predicted_slow_log_beta": -0.01005033585350145
        },
        {
          "seed": 2,
          "observed_mix_log_slope": -0.09078844601504571,
          "predicted_fast_log_beta": -0.10536051565782628,
          "predicted_slow_log_beta": -0.01005033585350145
        },
        {
          "seed": 3,
          "observed_mix_log_slope": -0.09078844601504561,
          "predicted_fast_log_beta": -0.10536051565782628,
          "predicted_slow_log_beta": -0.01005033585350145
        }
      ],
      "confirmed": true
    }
  },
  "custom_track": null
}