{ "bench_version": 1, "track": "dynamics", "model": "rnn_small", "metric_direction": "lower is better", "n_seeds": 8, "baseline": { "sweep": { "best_cfg": { "lr": 0.006 }, "sweep": [ { "cfg": { "lr": 0.001 }, "mean": 0.5709687098860741 }, { "cfg": { "lr": 0.003 }, "mean": 0.28142324835062027 }, { "cfg": { "lr": 0.006 }, "mean": 0.052330534905195236 } ], "full": { "mean": 0.04564603161998093, "std": 0.019217470172042173, "per_seed": [ 0.0358407162129879, 0.04454701393842697, 0.03819940611720085, 0.09073500335216522, 0.030078278854489326, 0.030246704816818237, 0.03592348471283913, 0.059597644954919815 ], "n": 8 } }, "best_config": { "lr": 0.006 }, "full": { "lr": 0.006, "periodic": false, "per_seed": [ 0.0358407162129879, 0.04454701393842697, 0.03819940611720085, 0.09073500335216522, 0.030078278854489326, 0.030246704816818237, 0.03592348471283913, 0.059597644954919815 ], "mean": 0.04564603161998093, "std": 0.020544339778379852 } }, "idea": { "lr": 0.006, "periodic": true, "per_seed": [ 0.03649410977959633, 0.04531322419643402, 0.03744680434465408, 0.08633020520210266, 0.030837101861834526, 0.02951119653880596, 0.03519521281123161, 0.062146708369255066 ], "mean": 0.04540932038798928, "std": 0.01951914347909807 }, "comparison": { "delta_mean": -0.00023671123199164867, "idea_wins": 4, "n_pairs": 8, "per_seed_diffs": [ 0.000653393566608429, 0.0007662102580070496, -0.0007526017725467682, -0.004404798150062561, 0.0007588230073451996, -0.0007355082780122757, -0.0007282719016075134, 0.002549063414335251 ], "p_value": 0.7555, "mde": 0.0016924713304113021, "mde_rel_pct": 3.707817022302648, "verdict": "no measurable effect", "system_worked": false }, "mechanism_signature": { "track_match": "dynamics is the built-in stability/control track", "schedule": { "phase_steps": [ 1, 1 ], "lr_multipliers": [ 1.6, 0.3999999999999999 ] }, "predicted_vs_observed": { "seed": 0, "test_mse": 0.03649410977959633, "predicted_rho_product": 0.9999960247634755, "observed_two_phase_gain": 1.000000023644702, "relative_error": 3.998881131933376e-06, "confirmed": true } }, "idea_sweep": [ { "lr": 0.006, "periodic": true, "per_seed": [ 0.03649410977959633, 0.04531322419643402, 0.03744680434465408, 0.08633020520210266, 0.030837101861834526, 0.02951119653880596, 0.03519521281123161, 0.062146708369255066 ], "mean": 0.04540932038798928, "std": 0.01951914347909807 }, { "lr": 0.001, "periodic": true, "per_seed": [ 0.4697585105895996, 0.6145305633544922, 0.5424293875694275, 0.6560655236244202, 0.49593931436538696, 0.5435400009155273, 0.5226424336433411, 0.6273190379142761 ], "mean": 0.5590280964970589, "std": 0.0664927567198825 }, { "lr": 0.003, "periodic": true, "per_seed": [ 0.22396183013916016, 0.29183024168014526, 0.2440265566110611, 0.3627980649471283, 0.21796903014183044, 0.23902076482772827, 0.23642562329769135, 0.3251190781593323 ], "mean": 0.26764389872550964, "std": 0.05300950027246798 } ], "runtime_sec": 87.40311336517334 }