{ "bench_version": 1, "track": "tangent_surface_vector_regression", "model": "shared_mlp", "metric_direction": "lower is better", "n_seeds": 8, "baseline": { "best_cfg": { "lr": 0.01 }, "sweep": [ { "cfg": { "lr": 0.001 }, "mean": 0.03764702659100294 }, { "cfg": { "lr": 0.003 }, "mean": 0.030783741734921932 }, { "cfg": { "lr": 0.01 }, "mean": 0.026880070567131042 } ], "union_full": [ { "cfg": { "lr": 0.001 }, "full": { "mean": 0.03800344513729215, "std": 0.0036767906806796174, "per_seed": [ 0.03472031280398369, 0.03560163080692291, 0.042614057660102844, 0.03765210509300232, 0.04322683811187744, 0.03853558748960495, 0.03991776704788208, 0.03175926208496094 ], "n": 8 } }, { "cfg": { "lr": 0.003 }, "full": { "mean": 0.03143041324801743, "std": 0.002817096624089596, "per_seed": [ 0.029847173020243645, 0.032056987285614014, 0.0286399032920599, 0.03259090334177017, 0.038180507719516754, 0.030082618817687035, 0.03025837428867817, 0.029786838218569756 ], "n": 8 } }, { "cfg": { "lr": 0.01 }, "full": { "mean": 0.02747795032337308, "std": 0.0043431611845433085, "per_seed": [ 0.02603250928223133, 0.02745499461889267, 0.026675675064325333, 0.027357103303074837, 0.038280390202999115, 0.02431103214621544, 0.0268161091953516, 0.02289578877389431 ], "n": 8 } } ], "full": { "mean": 0.02747795032337308, "std": 0.0043431611845433085, "per_seed": [ 0.02603250928223133, 0.02745499461889267, 0.026675675064325333, 0.027357103303074837, 0.038280390202999115, 0.02431103214621544, 0.0268161091953516, 0.02289578877389431 ], "n": 8 } }, "idea": { "mean": 0.021506645949557424, "std": 0.004870355712475741, "per_seed": [ 0.014883503317832947, 0.020688023418188095, 0.030457986518740654, 0.023641860112547874, 0.021189182996749878, 0.022041697055101395, 0.024708442389965057, 0.014442471787333488 ], "n": 8 }, "comparison": { "delta_mean": -0.005971304373815656, "idea_wins": 7, "n_pairs": 8, "per_seed_diffs": [ -0.011149005964398384, -0.006766971200704575, 0.0037823114544153214, -0.0037152431905269623, -0.017091207206249237, -0.002269335091114044, -0.0021076668053865433, -0.008453316986560822 ], "p_value": 0.03805, "mde": 0.005345683018986985, "mde_rel_pct": 19.45444604155894, "verdict": "idea better (significant)", "system_worked": true }, "custom_track": { "name": "tangent_surface_vector_regression", "file": "tangent_surface_track.py", "domain": "geometry/point-cloud" }, "idea_sweep": [ { "cfg": { "lr": 0.001 }, "full": { "mean": 0.02933011995628476, "std": 0.005050140438499371, "per_seed": [ 0.028748154640197754, 0.032053809612989426, 0.03960653766989708, 0.029504213482141495, 0.026794718578457832, 0.030350370332598686, 0.027094533666968346, 0.020488621667027473 ], "n": 8 } }, { "cfg": { "lr": 0.003 }, "full": { "mean": 0.024627438746392727, "std": 0.004374673952637387, "per_seed": [ 0.01918051764369011, 0.025951329618692398, 0.03225667402148247, 0.030034830793738365, 0.02338957041501999, 0.020098846405744553, 0.024956857785582542, 0.02115088328719139 ], "n": 8 } }, { "cfg": { "lr": 0.01 }, "full": { "mean": 0.021506645949557424, "std": 0.004870355712475741, "per_seed": [ 0.014883503317832947, 0.020688023418188095, 0.030457986518740654, 0.023641860112547874, 0.021189182996749878, 0.022041697055101395, 0.024708442389965057, 0.014442471787333488 ], "n": 8 } } ], "best_idea_cfg": { "lr": 0.01 }, "mechanism_signature": { "raw_mean_normal_component": 0.6702553331851959, "projected_max_normal_component": 4.76837158203125e-07, "rotation_relative_prediction_change": 0.25490777008235455, "predicted_projected_leakage": "<1e-5", "confirmed": true } }