{ "bench_version": 1, "track": "vision", "model": "directional_cnn", "metric_direction": "lower is better", "n_seeds": 8, "baseline": { "best_cfg": { "lr": 0.001 }, "sweep": [ { "cfg": { "lr": 0.001 }, "mean": 0.8287499845027924 }, { "cfg": { "lr": 0.003 }, "mean": 0.8387499749660492 }, { "cfg": { "lr": 0.005 }, "mean": 0.8549999892711639 } ], "full": { "mean": 0.8506249710917473, "std": 0.038522525390718286, "per_seed": [ 0.7999999523162842, 0.8849999904632568, 0.824999988079071, 0.7999999523162842, 0.8399999737739563, 0.85999995470047, 0.9099999666213989, 0.8849999904632568 ], "n": 8 } }, "idea": { "mean": 0.8393749892711639, "std": 0.03591983491614504, "per_seed": [ 0.8149999976158142, 0.8849999904632568, 0.8050000071525574, 0.7749999761581421, 0.8399999737739563, 0.8700000047683716, 0.875, 0.8499999642372131 ], "n": 8 }, "comparison": { "delta_mean": -0.011249981820583344, "idea_wins": 4, "n_pairs": 8, "per_seed_diffs": [ 0.01500004529953003, 0.0, -0.019999980926513672, -0.02499997615814209, 0.0, 0.010000050067901611, -0.034999966621398926, -0.0350000262260437 ], "p_value": 0.18755, "mde": 0.016685714835920434, "mde_rel_pct": 1.9615830010850617, "verdict": "no significant win", "system_worked": false }, "math_check": { "geometric_ratio": 0.5, "expected_ratio": 0.5, "ratio_allclose": true, "K": 8, "M": 2, "path_split_counts": [ 3, 3, 3, 3, 3, 3, 3, 3 ], "max_splits": 3, "N": 3, "bound_holds": true }, "idea_sweep": [ { "cfg": { "lr": 0.001 }, "result": { "mean": 0.8487499877810478, "std": 0.04846068349664526, "per_seed": [ 0.7649999856948853, 0.8799999952316284, 0.8849999904632568, 0.7849999666213989, 0.8199999928474426, 0.8700000047683716, 0.9049999713897705, 0.8799999952316284 ], "n": 8 } }, { "cfg": { "lr": 0.003 }, "result": { "mean": 0.8393749892711639, "std": 0.03591983491614504, "per_seed": [ 0.8149999976158142, 0.8849999904632568, 0.8050000071525574, 0.7749999761581421, 0.8399999737739563, 0.8700000047683716, 0.875, 0.8499999642372131 ], "n": 8 } }, { "cfg": { "lr": 0.005 }, "result": { "mean": 0.8443749845027924, "std": 0.05405075357696703, "per_seed": [ 0.8999999761581421, 0.7400000095367432, 0.8299999833106995, 0.8299999833106995, 0.7899999618530273, 0.8799999952316284, 0.8999999761581421, 0.8849999904632568 ], "n": 8 } } ], "custom_track": null, "structural_match": "spatial CIFAR images and convolutional directional fields", "mechanism_signature": { "prediction": "top-2 of K=8 uses 4x fewer directional samples", "K": 8, "beam": 2, "predicted_reduction": 4.0, "observed": [ { "seed": 0, "dense_metric": 0.7999999523162842, "routed_metric": 0.7649999856948853, "dense_samples_per_query": 12.0, "routed_samples_per_query": 3.0, "observed_reduction": 4.0, "routing_entropy": 2.0645053386688232 }, { "seed": 1, "dense_metric": 0.8799999952316284, "routed_metric": 0.8799999952316284, "dense_samples_per_query": 12.0, "routed_samples_per_query": 3.0, "observed_reduction": 4.0, "routing_entropy": 2.0756473541259766 } ], "confirmed": true } }