{ "bench_version": 1, "track": "tabular", "model": "fourier_mlp_shared", "metric_direction": "lower is better", "n_seeds": 8, "baseline": { "best_cfg": { "lr": 0.003 }, "sweep": [ { "cfg": { "lr": 0.001 }, "mean": 33.56439447402954 }, { "cfg": { "lr": 0.003 }, "mean": 22.75266695022583 }, { "cfg": { "lr": 0.01 }, "mean": 22.805188179016113 } ], "full": { "mean": 23.26982045173645, "std": 2.8220737062934167, "per_seed": [ 23.454627990722656, 20.045284271240234, 23.504892349243164, 24.005863189697266, 23.5848388671875, 27.756813049316406, 18.11393165588379, 25.692312240600586 ], "n": 8 } }, "idea": { "mean": 24.201251745224, "std": 1.9567082483775788, "per_seed": [ 25.041019439697266, 21.52345848083496, 25.802366256713867, 22.822996139526367, 21.128692626953125, 26.736886978149414, 25.048274993896484, 25.506319046020508 ], "n": 8 }, "comparison": { "delta_mean": 0.9314312934875488, "idea_wins": 4, "n_pairs": 8, "per_seed_diffs": [ 1.5863914489746094, 1.4781742095947266, 2.297473907470703, -1.1828670501708984, -2.456146240234375, -1.0199260711669922, 6.934343338012695, -0.18599319458007812 ], "p_value": 0.43755, "mde": 2.4404744399584257, "mde_rel_pct": 10.487723551714435, "verdict": "no significant win", "system_worked": false }, "notes": "Tabular is the matched built-in track for an input embedding/kernel feature intervention.", "mechanism_signature": { "prediction": "deterministic quadrature has lower top-10 RBF Gram eigenvalue error", "predicted_baseline_top10_relative_error": 1.4234343767166138, "predicted_idea_top10_relative_error": 4.320810317993164, "observed_delta_idea_minus_baseline": 2.8973759412765503, "confirmed": false } }