{
  "bench_version": 1,
  "track": "dynamics",
  "model": "rnn_small",
  "metric_direction": "lower is better",
  "n_seeds": 8,
  "baseline": {
    "best_cfg": {
      "lr": 0.006
    },
    "sweep": [
      {
        "cfg": {
          "lr": 0.001
        },
        "mean": 0.0019458123424556106
      },
      {
        "cfg": {
          "lr": 0.003
        },
        "mean": 0.000998065312160179
      },
      {
        "cfg": {
          "lr": 0.006
        },
        "mean": 0.0007849872272345237
      }
    ],
    "full": {
      "mean": 0.0007333773064601701,
      "std": 0.00023119481143299165,
      "per_seed": [
        0.0007082645315676928,
        0.0011445400305092335,
        0.00047928086132742465,
        0.0008078634855337441,
        0.00035395112354308367,
        0.0006696216878481209,
        0.0007750760996714234,
        0.0009284206316806376
      ],
      "n": 8
    }
  },
  "idea": {
    "mean": 0.0007333773064601701,
    "std": 0.00023119481143299165,
    "per_seed": [
      0.0007082645315676928,
      0.0011445400305092335,
      0.00047928086132742465,
      0.0008078634855337441,
      0.00035395112354308367,
      0.0006696216878481209,
      0.0007750760996714234,
      0.0009284206316806376
    ],
    "n": 8,
    "best_cfg": {
      "lr": 0.006,
      "lambda": 0.01
    }
  },
  "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
  },
  "idea_sweep": [
    {
      "cfg": {
        "lr": 0.001,
        "lambda": 0.01
      },
      "mean": 0.0019458123424556106
    },
    {
      "cfg": {
        "lr": 0.001,
        "lambda": 0.03
      },
      "mean": 0.0019458123424556106
    },
    {
      "cfg": {
        "lr": 0.001,
        "lambda": 0.1
      },
      "mean": 0.0019458123424556106
    },
    {
      "cfg": {
        "lr": 0.003,
        "lambda": 0.01
      },
      "mean": 0.000998065312160179
    },
    {
      "cfg": {
        "lr": 0.003,
        "lambda": 0.03
      },
      "mean": 0.000998065312160179
    },
    {
      "cfg": {
        "lr": 0.003,
        "lambda": 0.1
      },
      "mean": 0.000998065312160179
    },
    {
      "cfg": {
        "lr": 0.006,
        "lambda": 0.01
      },
      "mean": 0.0007849872272345237
    },
    {
      "cfg": {
        "lr": 0.006,
        "lambda": 0.03
      },
      "mean": 0.0007849872272345237
    },
    {
      "cfg": {
        "lr": 0.006,
        "lambda": 0.1
      },
      "mean": 0.0007849872272345237
    }
  ],
  "protocol_note": "Baseline and idea use identical rnn_small predictor, data, epochs, batch, Adam, and paired seeds; intervention is the Lyapunov loss.",
  "mechanism_signature": {
    "predicted_boundary": 0.0,
    "observed_mean_residual_eigenvalue": -0.4996017813682556,
    "observed_violation_fraction": 0.0,
    "observed_max_residual_eigenvalue": -0.38156095147132874,
    "alpha": 0.05,
    "h": 1.0,
    "confirmed": true,
    "signature_model_seed_metric": 0.0007082645315676928
  }
}