{
  "bench_version": 1,
  "track": "tabular",
  "model": "mlp_tiny",
  "metric_direction": "lower is better",
  "n_seeds": 8,
  "baseline": {
    "best_cfg": {
      "lr": 0.12,
      "epochs": 15
    },
    "sweep": [
      {
        "cfg": {
          "lr": 0.03,
          "epochs": 15
        },
        "mean": 17.87290906906128
      },
      {
        "cfg": {
          "lr": 0.06,
          "epochs": 15
        },
        "mean": 18.236158847808838
      },
      {
        "cfg": {
          "lr": 0.12,
          "epochs": 15
        },
        "mean": 7.929520010948181
      }
    ],
    "full": {
      "mean": 12.419063031673431,
      "std": 4.653164949772833,
      "per_seed": [
        7.232073783874512,
        9.054744720458984,
        8.681772232055664,
        6.7494893074035645,
        14.7591552734375,
        18.111841201782227,
        18.314743041992188,
        16.448684692382812
      ],
      "n": 8
    }
  },
  "idea": {
    "mean": 12.419063031673431,
    "std": 4.653164949772833,
    "per_seed": [
      7.232073783874512,
      9.054744720458984,
      8.681772232055664,
      6.7494893074035645,
      14.7591552734375,
      18.111841201782227,
      18.314743041992188,
      16.448684692382812
    ],
    "n": 8,
    "best_cfg": {
      "lr": 0.12,
      "epochs": 15
    },
    "sweep": [
      {
        "cfg": {
          "lr": 0.03,
          "epochs": 15
        },
        "mean": 22.84104561805725,
        "std": 11.555765903919722,
        "per_seed": [
          18.62317657470703,
          17.563692092895508,
          16.379343032836914,
          13.706111907958984,
          13.296585083007812,
          38.26602554321289,
          46.30095291137695,
          18.592477798461914
        ],
        "n": 8
      },
      {
        "cfg": {
          "lr": 0.06,
          "epochs": 15
        },
        "mean": 17.78099489212036,
        "std": 4.228547463617409,
        "per_seed": [
          9.133557319641113,
          17.798871994018555,
          16.796100616455078,
          15.525813102722168,
          24.372446060180664,
          17.10875129699707,
          21.30739402770996,
          20.20502471923828
        ],
        "n": 8
      },
      {
        "cfg": {
          "lr": 0.12,
          "epochs": 15
        },
        "mean": 12.419063031673431,
        "std": 4.653164949772833,
        "per_seed": [
          7.232073783874512,
          9.054744720458984,
          8.681772232055664,
          6.7494893074035645,
          14.7591552734375,
          18.111841201782227,
          18.314743041992188,
          16.448684692382812
        ],
        "n": 8,
        "best_cfg": {
          "lr": 0.12,
          "epochs": 15
        }
      }
    ]
  },
  "comparison": {
    "delta_mean": 0.0,
    "idea_wins": 0,
    "n_pairs": 8,
    "per_seed_diffs": [
      0.0,
      0.0,
      0.0,
      0.0,
      0.0,
      0.0,
      0.0,
      0.0
    ],
    "p_value": 1.0,
    "mde": 0.0,
    "mde_rel_pct": 0.0,
    "verdict": "no measurable effect",
    "system_worked": false
  },
  "mechanism_signature": {
    "prediction": "negative secant pairs are corrected to positive curvature; correction should occur only when the trained consecutive-gradient secant is negative",
    "baseline_observed": {
      "samples": 128,
      "mean_abs_parameter_input_gradient": 2.754227876663208,
      "finite_fraction": 1.0,
      "output_std": 3.702409267425537
    },
    "idea_observed": {
      "samples": 128,
      "mean_abs_parameter_input_gradient": 2.754227876663208,
      "finite_fraction": 1.0,
      "output_std": 3.702409267425537
    },
    "trained_pair_counts": {
      "baseline_negative_pairs": 0,
      "idea_negative_pairs": 0,
      "idea_corrected_pairs": 0
    },
    "confirmed": false,
    "measurement_note": "Behavior probes and pair counts are measured from trained official-tabular systems; no analytical toy identity is used."
  },
  "protocol_note": "Official registered tabular track; optimizer-only custom loop because the intervention changes training."
}