{ "bench_version": 1, "track": "tabular", "model": "mlp_tiny", "metric_direction": "lower is better", "n_seeds": 8, "baseline": { "best_cfg": { "lr": 0.006, "wd": 0.001 }, "sweep": [ { "cfg": { "lr": 0.001, "wd": 0.0 }, "mean": 17.79831099510193 }, { "cfg": { "lr": 0.001, "wd": 0.0001 }, "mean": 17.804412841796875 }, { "cfg": { "lr": 0.001, "wd": 0.001 }, "mean": 17.81385827064514 }, { "cfg": { "lr": 0.003, "wd": 0.0 }, "mean": 13.153547525405884 }, { "cfg": { "lr": 0.003, "wd": 0.0001 }, "mean": 13.136067628860474 }, { "cfg": { "lr": 0.003, "wd": 0.001 }, "mean": 13.114951848983765 }, { "cfg": { "lr": 0.006, "wd": 0.0 }, "mean": 9.905518293380737 }, { "cfg": { "lr": 0.006, "wd": 0.0001 }, "mean": 9.729963302612305 }, { "cfg": { "lr": 0.006, "wd": 0.001 }, "mean": 9.544940948486328 } ], "full": { "mean": 9.642327845096588, "std": 1.047326484601666, "per_seed": [ 9.891983985900879, 10.596479415893555, 8.670082092285156, 9.021218299865723, 7.615931034088135, 10.587333679199219, 9.936084747314453, 10.819509506225586 ], "n": 8 } }, "idea": { "mean": 9.909327685832977, "std": 1.2741297958180666, "per_seed": [ 10.186660766601562, 11.613065719604492, 8.699080467224121, 9.08471965789795, 7.6164374351501465, 10.533278465270996, 10.165548324584961, 11.37583065032959 ], "n": 8 }, "comparison": { "delta_mean": 0.26699984073638916, "idea_wins": 1, "n_pairs": 8, "per_seed_diffs": [ 0.2946767807006836, 1.0165863037109375, 0.028998374938964844, 0.06350135803222656, 0.0005064010620117188, -0.054055213928222656, 0.2294635772705078, 0.5563211441040039 ], "p_value": 0.04035, "mde": 0.30311716666322375, "mde_rel_pct": 3.143609837093103, "verdict": "idea worse (significant)", "system_worked": false }, "mechanism_signature": { "track_choice": "tabular matches optimizer/regularizer ideas; shared MLP.", "idea_grid": [ { "cfg": { "lr": 0.001, "eps0": 0.0001, "eps_min": 1e-06, "decay": 0.7 }, "result": { "mean": 18.25679302215576, "std": 1.6583796998792717, "per_seed": [ 17.28780174255371, 17.48930549621582, 20.83658218383789, 15.395792007446289, 17.73833656311035, 17.7778263092041, 19.221439361572266, 20.307260513305664 ], "n": 8 } }, { "cfg": { "lr": 0.003, "eps0": 0.001, "eps_min": 1e-06, "decay": 0.7 }, "result": { "mean": 13.504453182220459, "std": 1.1547149957425087, "per_seed": [ 13.25252914428711, 13.844292640686035, 13.66245174407959, 11.70921802520752, 12.26156234741211, 13.167490005493164, 14.4833984375, 15.654683113098145 ], "n": 8 } }, { "cfg": { "lr": 0.006, "eps0": 0.01, "eps_min": 1e-06, "decay": 0.7 }, "result": { "mean": 9.909327685832977, "std": 1.2741297958180666, "per_seed": [ 10.186660766601562, 11.613065719604492, 8.699080467224121, 9.08471965789795, 7.6164374351501465, 10.533278465270996, 10.165548324584961, 11.37583065032959 ], "n": 8 } } ], "selected_idea_cfg": { "lr": 0.006, "eps0": 0.01, "eps_min": 1e-06, "decay": 0.7 }, "signature": { "prediction": "damped CG residual <= 1e-5 max(1, ||b||) on trained network", "predicted_max_residual": 0.0005065035629272461, "observed_max_residual": 80.76493072509766, "epsilon_pair_relative_change_norm_proxy": 0.0031547897960990667, "trained_model_parameter_norm": 8.515052795410156, "confirmed": false } } }