{ "bench_version": 1, "track": "multihorizon_sequence", "model": "custom_gru_decoder", "metric_direction": "lower is better", "n_seeds": 8, "baseline": { "best_cfg": { "lr": 0.006, "scale": 1.0 }, "sweep": [ { "cfg": { "lr": 0.001, "scale": 0.5 }, "mean": 0.36861784756183624 }, { "cfg": { "lr": 0.001, "scale": 1.0 }, "mean": 0.3673807680606842 }, { "cfg": { "lr": 0.003, "scale": 0.5 }, "mean": 0.26081548631191254 }, { "cfg": { "lr": 0.003, "scale": 1.0 }, "mean": 0.2626522071659565 }, { "cfg": { "lr": 0.006, "scale": 0.5 }, "mean": 0.2287113033235073 }, { "cfg": { "lr": 0.006, "scale": 1.0 }, "mean": 0.22365260869264603 } ], "full": { "mean": 0.2238331437110901, "std": 0.021044464376645176, "per_seed": [ 0.2033109813928604, 0.24181047081947327, 0.23719552159309387, 0.21229346096515656, 0.2200879603624344, 0.26243942975997925, 0.220900297164917, 0.19262702763080597 ], "n": 8 } }, "idea": { "mean": 0.207510719075799, "std": 0.016642018572369865, "per_seed": [ 0.20119042694568634, 0.23540125787258148, 0.23484113812446594, 0.2042408287525177, 0.20302963256835938, 0.1936691254377365, 0.19913142919540405, 0.1885819137096405 ], "n": 8 }, "comparison": { "delta_mean": -0.0163224246352911, "idea_wins": 8, "n_pairs": 8, "per_seed_diffs": [ -0.0021205544471740723, -0.006409212946891785, -0.0023543834686279297, -0.008052632212638855, -0.017058327794075012, -0.06877030432224274, -0.02176886796951294, -0.004045113921165466 ], "p_value": 0.0081, "mde": 0.018684083393322008, "mde_rel_pct": 8.347326532409454, "verdict": "idea better (significant)", "system_worked": true }, "mechanism_signature": { "signature": { "prediction": "reverse cumulative shell makes early outputs more sensitive to later coefficients", "predicted": { "early_over_late_sensitivity": ">1" }, "observed": { "early_abs_change": 0.0, "late_abs_change": 0.016314398497343063, "early_over_late_sensitivity": 0.0, "trained_test_mse": 0.20119042694568634 }, "confirmed": false, "forward_seconds": 1.733663829996658 }, "idea_sweep": [ { "cfg": { "lr": 0.006, "scale": 1.0 }, "result": { "mean": 0.207510719075799, "std": 0.016642018572369865, "per_seed": [ 0.20119042694568634, 0.23540125787258148, 0.23484113812446594, 0.2042408287525177, 0.20302963256835938, 0.1936691254377365, 0.19913142919540405, 0.1885819137096405 ], "n": 8 } }, { "cfg": { "lr": 0.001, "scale": 1.0 }, "result": { "mean": 0.339939147233963, "std": 0.038412014687187156, "per_seed": [ 0.35680338740348816, 0.38626959919929504, 0.4020107686519623, 0.31106022000312805, 0.35761505365371704, 0.29985377192497253, 0.2990495562553406, 0.3068508207798004 ], "n": 8 } }, { "cfg": { "lr": 0.006, "scale": 0.5 }, "result": { "mean": 0.21021396107971668, "std": 0.022721494928976002, "per_seed": [ 0.21364261209964752, 0.23506788909435272, 0.23733554780483246, 0.20955374836921692, 0.2235165685415268, 0.21272434294223785, 0.1817266047000885, 0.1681443750858307 ], "n": 8 } } ], "complexity_prediction": { "shell": "O(Kd)", "dense": "O(K^2d)" }, "custom_track": { "name": "multihorizon_sequence", "file": "custom_track.py", "domain": "sequence" } } }