{
  "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": 0.0002686133266252
      },
      {
        "cfg": {
          "lr": 0.003
        },
        "mean": 2.2429956516134553e-05
      },
      {
        "cfg": {
          "lr": 0.006
        },
        "mean": 1.2719097412627889e-05
      }
    ],
    "full": {
      "mean": 1.3684802866009704e-05,
      "std": 2.4297331089709535e-06,
      "per_seed": [
        1.0965275578200817e-05,
        1.3063688129477669e-05,
        1.363205592497252e-05,
        1.321537001786055e-05,
        1.694267484708689e-05,
        1.790331407391932e-05,
        1.3341849808057304e-05,
        1.0414194548502564e-05
      ],
      "n": 8
    }
  },
  "idea": {
    "mean": 1.1164625242088277e-06,
    "std": 6.736045713192092e-07,
    "per_seed": [
      2.8550934416671225e-07,
      2.1125936200405704e-06,
      1.6401930906795315e-06,
      3.9009708530102216e-07,
      1.0302179589416482e-06,
      1.806109253266186e-06,
      1.334532043983927e-06,
      3.3244779729102447e-07
    ],
    "n": 8,
    "chosen_cfg": {
      "lr": 0.006,
      "tmax": 8,
      "alpha": 0.25,
      "halt_bias": 0.0
    },
    "sweep": [
      {
        "cfg": {
          "lr": 0.006,
          "tmax": 8,
          "alpha": 0.25,
          "halt_bias": 0.0
        },
        "mean": 1.107098285046959e-06,
        "per_seed": [
          2.8550934416671225e-07,
          2.1125936200405704e-06,
          1.6401930906795315e-06,
          3.9009708530102216e-07
        ]
      },
      {
        "cfg": {
          "lr": 0.001,
          "tmax": 8,
          "alpha": 0.25,
          "halt_bias": 0.0
        },
        "mean": 5.165409334040305e-05,
        "per_seed": [
          1.0778670002764557e-05,
          8.488429011777043e-05,
          8.795395115157589e-05,
          2.299946208950132e-05
        ]
      },
      {
        "cfg": {
          "lr": 0.006,
          "tmax": 8,
          "alpha": 0.25,
          "halt_bias": 0.0
        },
        "mean": 1.107098285046959e-06,
        "per_seed": [
          2.8550934416671225e-07,
          2.1125936200405704e-06,
          1.6401930906795315e-06,
          3.9009708530102216e-07
        ]
      }
    ]
  },
  "comparison": {
    "delta_mean": -1.2568340341800877e-05,
    "idea_wins": 8,
    "n_pairs": 8,
    "per_seed_diffs": [
      -1.0679766234034105e-05,
      -1.0951094509437098e-05,
      -1.1991862834292988e-05,
      -1.2825272932559528e-05,
      -1.5912456888145243e-05,
      -1.6097204820653133e-05,
      -1.2007317764073377e-05,
      -1.008174675121154e-05
    ],
    "p_value": 0.0081,
    "mde": 1.9149415921520303e-06,
    "mde_rel_pct": 13.993198228001951,
    "verdict": "idea better (significant)",
    "system_worked": true
  },
  "mechanism_signature": {
    "prediction": "adaptive recurrence reduces average executed microsteps as halting bias increases, while sharing one transition rule",
    "predicted": {
      "higher_bias": "lower_steps"
    },
    "observed": {
      "idea_mean_weighted_steps": 1.287009134888649,
      "idea_mean_halt_mass": 1.0,
      "all_steps": [
        1.235519289970398,
        1.3300005197525024,
        1.2526575326919556,
        1.3078808784484863,
        1.3313579559326172,
        1.2955666780471802,
        1.2435346841812134,
        1.2995555400848389
      ]
    },
    "confirmed": true,
    "parameter_counts": {
      "baseline": 13313,
      "idea": 17090
    }
  },
  "custom_track": null
}