Descent-Certified LMO Sign Switching / report_bench_2026-08-31T194227.md
Failed on benchmark
Стенд-проверка (stage-2) · промт оператора:
(универсальный)
Ответ агента:
{ "worked": false, "confidence": 9, "verdict": "Implemented Descent-Certified LMO Sign Switching as a local optimizer training loop for the structurally appropriate tabular optimizer track, with post-LMO baseline, pre-LMO residual feedback, independent held-out minibatch alignment, and matched hyperparameter sweeps. Across 8 paired seeds, the best idea configuration had MSE 21.5859 versus baseline 21.5750, paired delta +0.0109 and permutation p=0.8426, so there was no significant improvement. The trained-model mechanism signature was confirmed, but it did not translate into a task-metric win.", "metrics": { "baseline": "tabular/mlp_tiny post-LMO SignMuon, best lr=0.006 beta=0.9: mean test MSE=21.575043, std=1.915759, per-seed=[22.023918,17.909542,21.564068,20.560976,20.688057,23.367746,21.641396,24.844645]", "idea": "Descent-certified switch, best lr=0.006 beta=0.9 tau=0.0: mean test MSE=21.585942, std=1.884198, per-seed=[22.238335,17.963984,21.634672,20.441940,20.853069,23.095635,21.598305,24.861594]; delta=+0.010898, p=0.8426, wins=3/8; mechanism confirmed with post_fraction=0.960648, predicted_fraction=0.960648, post_negative=0.039352, chosen_negative=0.033069" }, "bench_report": { "bench_version": 1, "track": "tabular", "model": "mlp_tiny", "metric_direction": "lower is better", "n_seeds": 8, "baseline": { "best_cfg": { "lr": 0.006, "beta": 0.9 }, "sweep_configs": "lr in [0.001,0.003,0.006], beta in [0.8,0.9]; all idea learning rates included", "full": { "mean": 21.575043439865112, "std": 1.9157591398399918, "per_seed": [ 22.02391815185547, 17.909542083740234, 21.564067840576172, 20.560976028442383, 20.68805694580078, 23.367746353149414, 21.641395568847656, 24.84464454650879 ], "n": 8 } }, "idea": { "best_cfg": { "lr": 0.006, "beta": 0.9, "tau": 0.0 }, "full": { "mean": 21.585941553115845, "std": 1.8841984845237936, "per_seed": [ 22.23833465576172, 17.9639835357666, 21.634672164916992, 20.441940307617188, 20.853069305419922, 23.09563446044922, 21.598304748535156, 24.86159324645996 ], "n": 8 }, "sweep_taus": [ 0.0, 0.05, 0.1 ] }, "comparison": { "delta_mean": 0.010898113250732422, "idea_wins": 3, "n_pairs": 8, "per_seed_diffs": [ 0.21441650390625, 0.05444145202636719, 0.07060432434082031, -0.11903572082519531, 0.16501235961914062, -0.2721118927001953, -0.0430908203125, 0.016948699951171875 ], "p_value": 0.8426, "verdict": "no measurable effect", "system_worked": false }, "mechanism_signature": { "prediction": "held-out alignment gating routes post when rho>=tau and reduces negative held-out alignment", "observed": { "post_fraction": 0.9606481481481481, "predicted_fraction": 0.9606481481481481, "post_negative": 0.039351851851851846, "chosen_negative": 0.03306878306878307, "rho_mean": 0.2859362520431751, "confirmed": true }, "trained_models": true }, "custom_track": null }, "how_to_run": "/home/maxwelhelp/main/bin/python3 bench_switch.py", "files": [ "bench_switch.py", "bench_report.json" ], "limitations": "Only the structurally appropriate built-in tabular track was tested, not CIFAR, sequence, or dynamics. Runtime and communication overhead were not separately benchmarked; the implementation uses a compact custom SignMuon-style loop rather than the bench Adam train_model path because the idea changes the optimizer itself.", "system_verdict": "failed", "practical_verdict": "no_effect", "mechanism_ok": 0, "system_judged": true }