{
  "bench_version": 1,
  "track": "manifold_pendulum",
  "model": "rnn_small",
  "metric_direction": "lower is better",
  "n_seeds": 8,
  "baseline": {
    "best_cfg": {
      "lr": 0.01
    },
    "sweep": [
      {
        "cfg": {
          "lr": 0.001
        },
        "mean": 0.30809812992811203
      },
      {
        "cfg": {
          "lr": 0.003
        },
        "mean": 0.04099804675206542
      },
      {
        "cfg": {
          "lr": 0.01
        },
        "mean": 0.007313748239539564
      }
    ],
    "full": {
      "mean": 0.0062498191255144775,
      "std": 0.001496702937330929,
      "per_seed": [
        0.00792852696031332,
        0.00787876546382904,
        0.0054243444465100765,
        0.008023356087505817,
        0.00436387537047267,
        0.006577811203896999,
        0.004088760819286108,
        0.005713112652301788
      ],
      "n": 8
    }
  },
  "idea": {
    "mean": 0.017734745517373085,
    "std": 0.0075728509012216545,
    "per_seed": [
      0.024637481197714806,
      0.015422593802213669,
      0.03352177515625954,
      0.011271449737250805,
      0.01013706810772419,
      0.020766805857419968,
      0.011616603471338749,
      0.014504186809062958
    ],
    "n": 8
  },
  "comparison": {
    "delta_mean": 0.011484926391858608,
    "idea_wins": 0,
    "n_pairs": 8,
    "per_seed_diffs": [
      0.016708954237401485,
      0.007543828338384628,
      0.02809743070974946,
      0.0032480936497449875,
      0.00577319273725152,
      0.014188994653522968,
      0.007527842652052641,
      0.00879107415676117
    ],
    "p_value": 0.0081,
    "mde": 0.006700379570878075,
    "mde_rel_pct": 107.20917575877058,
    "verdict": "idea worse (significant)",
    "system_worked": false
  },
  "mechanism_signature": {
    "custom_track": {
      "name": "manifold_pendulum",
      "file": "manifold_dynamics_track.py",
      "domain": "dynamics_and_embedded_manifolds"
    },
    "idea_config": {
      "lr": 0.01
    },
    "idea_sweep": [
      {
        "cfg": {
          "lr": 0.001
        },
        "result": {
          "mean": 0.26221400313079357,
          "std": 0.05336933195312093,
          "per_seed": [
            0.352046400308609,
            0.3472537696361542,
            0.249238058924675,
            0.23107391595840454,
            0.24011655151844025,
            0.25662562251091003,
            0.19462718069553375,
            0.22673052549362183
          ],
          "n": 8
        }
      },
      {
        "cfg": {
          "lr": 0.003
        },
        "result": {
          "mean": 0.08073520846664906,
          "std": 0.02050061232661446,
          "per_seed": [
            0.11787223815917969,
            0.08001995086669922,
            0.1040424108505249,
            0.06887325644493103,
            0.0729207918047905,
            0.08841918408870697,
            0.06128842756152153,
            0.05244540795683861
          ],
          "n": 8
        }
      },
      {
        "cfg": {
          "lr": 0.01
        },
        "result": {
          "mean": 0.017734745517373085,
          "std": 0.0075728509012216545,
          "per_seed": [
            0.024637481197714806,
            0.015422593802213669,
            0.03352177515625954,
            0.011271449737250805,
            0.01013706810772419,
            0.020766805857419968,
            0.011616603471338749,
            0.014504186809062958
          ],
          "n": 8
        }
      }
    ],
    "mechanism_signature": {
      "prediction": "intrinsic retraction keeps every predicted embedded angular state on S1; Euclidean output has nonzero norm error",
      "predicted_baseline_violation_order": "nonzero",
      "predicted_idea_violation": 0.0,
      "observed_baseline_max_violation": 0.16893672943115234,
      "observed_idea_max_violation": 1.1920928955078125e-07,
      "observed_idea_metric_seed0": 0.024637481197714806,
      "confirmed": true
    }
  }
}