Spectral-Gated Parallel Best Responses / report_bench_2026-09-02T001805.md
Failed on benchmark
Стенд-проверка (stage-2) · промт оператора:
(универсальный)
Ответ агента:
{ "worked": false, "confidence": 9, "verdict": "Implemented Spectral-Gated Parallel Best Responses as a blockwise optimizer on the structurally matched tabular Friedman#1 track with mlp_tiny. Across 8 paired seeds, tuned Adam achieved test MSE 10.6786 while the best idea configuration achieved 98.1757; paired delta was +87.4970 with permutation p=0.0081, so the idea was significantly worse. The NN-scale heuristic mechanism signature was not confirmed.", "metrics": { "baseline": "Tuned Adam, lr=0.006 and weight_decay=0.0001: mean test MSE 10.6786, std 1.2287.", "idea": "Best gated blockwise update, lr=0.006, reg=0.02, alpha=0.5: mean test MSE 98.1757, std 15.0264; delta +87.4970; p=0.0081; 0/8 wins." }, "how_to_run": "/home/maxwelhelp/main/bin/python3 bench_experiment.py", "files": [ "bench_experiment.py", "bench_report.json" ], "limitations": "Only tabular/Friedman#1 with mlp_tiny was tested. The neural coupling estimate was a gradient-norm heuristic rather than exact Hessian-vector-product spectral estimation. Runtime/FLOP overhead, larger models, and the other benchmark tracks were not tested.", "bench_report": { "bench_version": 1, "track": "tabular", "model": "mlp_tiny", "metric_direction": "lower is better", "n_seeds": 8, "baseline": { "best_cfg": { "lr": 0.006, "weight_decay": 0.0001 }, "full": { "mean": 10.678602695465088, "std": 1.2286671921700971, "per_seed": [ 10.056567192077637, 11.58691120147705, 9.562552452087402, 9.74867057800293, 8.789999961853027, 11.738417625427246, 11.37514591217041, 12.570556640625 ], "n": 8 }, "sweep_configs": 6 }, "idea": { "cfg": { "lr": 0.006, "reg": 0.02, "alpha": 0.5 }, "mean": 98.17565250396729, "std": 15.026422279731827, "per_seed": [ 108.2091293334961, 109.38567352294922, 88.7021255493164, 95.20142364501953, 67.35282135009766, 112.1347427368164, 89.88146209716797, 114.537841796875 ], "n": 8 }, "comparison": { "delta_mean": 87.4970498085022, "idea_wins": 0, "n_pairs": 8, "p_value": 0.0081, "verdict": "idea worse (significant)", "system_worked": false }, "mechanism_signature": { "confirmed": false, "prediction": "gated blocks should avoid unstable coupling", "note": "NN probe was heuristic, not exact Hessian spectral radius" }, "custom_track": null }, "system_verdict": "failed", "practical_verdict": "harms", "mechanism_ok": 0, "system_judged": true }