{ "bench_version": 1, "track": "dynamics", "model": "shared_gru_latent_linear_head", "metric_direction": "lower is better", "n_seeds": 8, "baseline": { "best_cfg": { "lr": 0.01 }, "sweep": [ { "cfg": { "lr": 0.001 }, "mean": 0.8660691231489182 }, { "cfg": { "lr": 0.003 }, "mean": 0.16844721138477325 }, { "cfg": { "lr": 0.01 }, "mean": 0.030930949840694666 } ], "full": { "mean": 0.03257687599398196, "std": 0.00921168354607454, "per_seed": [ 0.027142083272337914, 0.02880755066871643, 0.039878763258457184, 0.027895402163267136, 0.050688546150922775, 0.021972693502902985, 0.024402838200330734, 0.0398271307349205 ], "n": 8 } }, "idea": { "mean": 0.0003928114747395739, "std": 0.00021742121198405602, "per_seed": [ 8.355393947567791e-05, 0.0004895747406408191, 0.0001629032049095258, 0.0004371287068352103, 0.0004481615615077317, 0.0004002868663519621, 0.0008424268453381956, 0.0002784559328574687 ], "n": 8 }, "comparison": { "delta_mean": -0.032184064519242384, "idea_wins": 8, "n_pairs": 8, "per_seed_diffs": [ -0.027058529332862236, -0.02831797592807561, -0.03971586005354766, -0.027458273456431925, -0.050240384589415044, -0.021572406636551023, -0.02356041135499254, -0.03954867480206303 ], "p_value": 0.0081, "mde": 0.008286675296304802, "mde_rel_pct": 25.437292691403652, "verdict": "idea better (significant)", "system_worked": true }, "mechanism_signature": { "idea_sweep": [ { "cfg": { "lr": 0.001 }, "mean": 0.00040330794945475645, "full": { "mean": 0.00040330794945475645, "std": 0.0001870869897132185, "per_seed": [ 0.00025814544642344117, 0.00038167217280715704, 0.0006836550310254097, 0.0006370171322487295, 0.00021974655101075768, 0.00024331139866262674, 0.0005830578738823533, 0.00021985798957757652 ], "n": 8 } }, { "cfg": { "lr": 0.003 }, "mean": 0.0004725444596260786, "full": { "mean": 0.0004725444596260786, "std": 0.00012726428121603695, "per_seed": [ 0.0003564116486813873, 0.000323356653098017, 0.0004201834963168949, 0.0007025530212558806, 0.00039990260847844183, 0.0004050004936289042, 0.0006209795828908682, 0.0005519681726582348 ], "n": 8 } }, { "cfg": { "lr": 0.01 }, "mean": 0.0003928114747395739, "full": { "mean": 0.0003928114747395739, "std": 0.00021742121198405602, "per_seed": [ 8.355393947567791e-05, 0.0004895747406408191, 0.0001629032049095258, 0.0004371287068352103, 0.0004481615615077317, 0.0004002868663519621, 0.0008424268453381956, 0.0002784559328574687 ], "n": 8 } } ], "paired_delta": { "per_seed": [ -0.027058529332862236, -0.02831797592807561, -0.03971586005354766, -0.027458273456431925, -0.050240384589415044, -0.021572406636551023, -0.02356041135499254, -0.03954867480206303 ], "mean": -0.032184064519242384, "permutation_p": 0.0081 }, "mechanism_signature": { "near_predictive_variance": 0.0006225919933058321, "far_predictive_variance": 0.0006237539928406477, "observed_far_near_ratio": 1.0018663901034859, "predicted_far_near_ratio": 1.0018666982650757, "interval_90pct_coverage": 1.0, "confirmed": true }, "structural_match": "controlled damped pendulum multi-step dynamics" } }