{ "bench_version": 1, "track": "tabular", "model": "mlp_tiny", "metric_direction": "lower is better", "n_seeds": 8, "baseline": { "best_cfg": { "lr": 0.03, "betas": [ 0.9, 0.999 ] }, "sweep": [ { "cfg": { "lr": 0.003, "betas": [ 0.9, 0.999 ] }, "mean": 193.97412109375 }, { "cfg": { "lr": 0.003, "betas": [ 0.9, 0.99 ] }, "mean": 197.73908233642578 }, { "cfg": { "lr": 0.01, "betas": [ 0.9, 0.999 ] }, "mean": 17.7281391620636 }, { "cfg": { "lr": 0.01, "betas": [ 0.9, 0.99 ] }, "mean": 17.624634265899658 }, { "cfg": { "lr": 0.03, "betas": [ 0.9, 0.999 ] }, "mean": 9.181843519210815 }, { "cfg": { "lr": 0.03, "betas": [ 0.9, 0.99 ] }, "mean": 9.269299268722534 } ], "full": { "mean": 8.84202092885971, "std": 1.0786474804404615, "per_seed": [ 10.21053409576416, 7.164846420288086, 10.157970428466797, 9.194023132324219, 7.742873668670654, 9.739964485168457, 8.464797019958496, 8.061158180236816 ], "n": 8 } }, "idea": { "mean": 155.1243715286255, "std": 25.433554864851153, "per_seed": [ 171.2412567138672, 187.00245666503906, 195.0430908203125, 146.14537048339844, 144.15423583984375, 147.81997680664062, 114.51549530029297, 135.07308959960938 ], "n": 8, "tested_lr_candidates": [ 0.003, 0.01, 0.03 ] }, "comparison": { "delta_mean": 146.28235059976578, "idea_wins": 0, "n_pairs": 8, "per_seed_diffs": [ 161.03072261810303, 179.83761024475098, 184.8851203918457, 136.95134735107422, 136.4113621711731, 138.08001232147217, 106.05069828033447, 127.01193141937256 ], "p_value": 0.0081, "mde": 22.514930098157226, "mde_rel_pct": 254.63556667989934, "verdict": "idea worse (significant)", "system_worked": false }, "mechanism_signature": { "prediction": "QR projector splitting preserves orthogonality and avoids inverse-S instability", "trained_model_observed": { "chosen_lr": 0.03, "u_orth_max": 6.055437324903323e-07, "v_orth_max": 5.146836201674887e-07, "sigma_min_range": [ 0.15457221865653992, 0.24608534574508667 ], "factor_norm_max": 18.993122100830078, "finite_all": true }, "confirmed": true } }