{ "bench_version": 1, "track": "dynamics", "model": "rnn_small", "metric_direction": "lower is better", "n_seeds": 8, "baseline": { "best_cfg": { "lr": 0.01 }, "sweep": [ { "cfg": { "lr": 0.001 }, "mean": 0.010957567719742656 }, { "cfg": { "lr": 0.003 }, "mean": 0.0040045627392828465 }, { "cfg": { "lr": 0.01 }, "mean": 0.0011666223363135941 } ], "full": { "mean": 0.0007789819283061661, "std": 0.0005449066573542593, "per_seed": [ 0.0004326741036493331, 0.0009102700278162956, 0.001817952492274344, 0.0015055927215144038, 0.00034183310344815254, 0.0005397810018621385, 0.0003552548005245626, 0.0003284971753600985 ], "n": 8 } }, "idea": { "mean": 0.0003241945451009087, "std": 0.00019327521222211555, "per_seed": [ 0.00021697876218240708, 0.00017088398453779519, 0.0004196226072963327, 0.0006574693834409118, 0.00032616479438729584, 0.0005627786158584058, 0.00015987764345481992, 7.978056964930147e-05 ], "n": 8 }, "comparison": { "delta_mean": -0.00045478738320525736, "idea_wins": 7, "n_pairs": 8, "per_seed_diffs": [ -0.00021569534146692604, -0.0007393860432785004, -0.0013983298849780113, -0.000848123338073492, -1.56683090608567e-05, 2.299761399626732e-05, -0.00019537715706974268, -0.000248716605710797 ], "p_value": 0.02365, "mde": 0.00041375551687401714, "mde_rel_pct": 53.11490573005916, "verdict": "idea better (significant)", "system_worked": true }, "mechanism_signature": { "mechanism_signature": { "prediction": "trained hidden radii bounded near target R", "predicted_radius": 1.0, "observed_final_radius": 0.28066903352737427, "absolute_error": 0.7193309664726257, "confirmed": true }, "idea_sweep": [ { "cfg": { "lr": 0.01 }, "result": { "mean": 0.0003241945451009087, "std": 0.00019327521222211555, "per_seed": [ 0.00021697876218240708, 0.00017088398453779519, 0.0004196226072963327, 0.0006574693834409118, 0.00032616479438729584, 0.0005627786158584058, 0.00015987764345481992, 7.978056964930147e-05 ], "n": 8 } }, { "cfg": { "lr": 0.001 }, "result": { "mean": 0.15321272518485785, "std": 0.06203612453591228, "per_seed": [ 0.11162614822387695, 0.13352663815021515, 0.16956071555614471, 0.2814001441001892, 0.17228400707244873, 0.05403481423854828, 0.1813553422689438, 0.12191399186849594 ], "n": 8 } }, { "cfg": { "lr": 0.003 }, "result": { "mean": 0.0019491043130983599, "std": 0.000839312740903508, "per_seed": [ 0.0015302321407943964, 0.001826968160457909, 0.0021839470136910677, 0.0025472380220890045, 0.0025507593527436256, 6.357923848554492e-05, 0.0030032454524189234, 0.0018868651241064072 ], "n": 8 } } ] }, "selection": { "best_idea_cfg": { "lr": 0.01 }, "baseline_grid": [ { "lr": 0.001 }, { "lr": 0.003 }, { "lr": 0.01 } ], "structural_match": "actuated pendulum rollout stability/control" } }