{
  "bench_version": 1,
  "track": "tabular",
  "model": "mlp_tiny",
  "metric_direction": "lower is better",
  "n_seeds": 8,
  "baseline": {
    "best_cfg": {
      "lr": 0.003,
      "weight_decay": 0.0
    },
    "sweep": [
      {
        "cfg": {
          "lr": 0.001,
          "weight_decay": 0.0
        },
        "mean": 188.35083389282227
      },
      {
        "cfg": {
          "lr": 0.003,
          "weight_decay": 0.0
        },
        "mean": 19.947585344314575
      },
      {
        "cfg": {
          "lr": 0.006,
          "weight_decay": 0.0
        },
        "mean": 22.523260593414307
      }
    ],
    "full": {
      "mean": 21.810271620750427,
      "std": 6.031322024802775,
      "per_seed": [
        17.455322265625,
        15.50521183013916,
        29.37735366821289,
        17.45245361328125,
        33.27134323120117,
        16.96589469909668,
        22.887645721435547,
        21.56694793701172
      ],
      "n": 8
    }
  },
  "idea": {
    "mean": 21.595937252044678,
    "std": 5.868687330662625,
    "per_seed": [
      17.42387580871582,
      15.323787689208984,
      28.447860717773438,
      17.114221572875977,
      32.90999221801758,
      16.928159713745117,
      22.993961334228516,
      21.625638961791992
    ],
    "n": 8
  },
  "comparison": {
    "delta_mean": -0.2143343687057495,
    "idea_wins": 6,
    "n_pairs": 8,
    "per_seed_diffs": [
      -0.03144645690917969,
      -0.18142414093017578,
      -0.9294929504394531,
      -0.33823204040527344,
      -0.36135101318359375,
      -0.0377349853515625,
      0.10631561279296875,
      0.05869102478027344
    ],
    "p_value": 0.09385,
    "mde": 0.2811647999151475,
    "mde_rel_pct": 1.2891393780151073,
    "verdict": "no measurable effect",
    "system_worked": false
  },
  "mechanism_signature": {
    "prediction": {
      "feature_dim": 64,
      "observed_rank_last": 64,
      "predicted_identity_max_eigen_mean": 1.8750072717666626,
      "observed_QtQ_error_max": 1.0,
      "prediction": "MGS memory predicts identity Gramian after d independent directions",
      "confirmed": false
    },
    "idea_configs": [
      {
        "cfg": {
          "lr": 0.003,
          "lambda": 0.02
        },
        "result": {
          "mean": 21.754212260246277,
          "std": 5.9932442203746445,
          "per_seed": [
            17.445890426635742,
            15.479540824890137,
            29.154930114746094,
            17.36090087890625,
            33.19727325439453,
            16.9453067779541,
            22.87087631225586,
            21.5789794921875
          ],
          "n": 8
        }
      },
      {
        "cfg": {
          "lr": 0.003,
          "lambda": 0.05
        },
        "result": {
          "mean": 21.67228603363037,
          "std": 5.945718685399587,
          "per_seed": [
            17.41834259033203,
            15.36572265625,
            28.8680419921875,
            17.256317138671875,
            33.06776428222656,
            16.93799591064453,
            22.85132598876953,
            21.612777709960938
          ],
          "n": 8
        }
      },
      {
        "cfg": {
          "lr": 0.003,
          "lambda": 0.1
        },
        "result": {
          "mean": 21.595937252044678,
          "std": 5.868687330662625,
          "per_seed": [
            17.42387580871582,
            15.323787689208984,
            28.447860717773438,
            17.114221572875977,
            32.90999221801758,
            16.928159713745117,
            22.993961334228516,
            21.625638961791992
          ],
          "n": 8
        }
      }
    ],
    "confirmed": false
  },
  "track_justification": "Tabular is the prescribed optimizer track; Friedman regression with mlp_tiny has a trainable final linear head and fixed feature dimension, matching the memory correction."
}