{ "bench_version": 1, "track": "tabular", "model": "mlp_tiny", "metric_direction": "lower is better", "n_seeds": 8, "baseline": { "sweep": [ { "cfg": { "lr": 0.001, "wd": 0.0 }, "mean": 17.79831099510193 }, { "cfg": { "lr": 0.001, "wd": 0.0001 }, "mean": 17.804412841796875 }, { "cfg": { "lr": 0.003, "wd": 0.0 }, "mean": 13.153547525405884 }, { "cfg": { "lr": 0.003, "wd": 0.0001 }, "mean": 13.136067628860474 }, { "cfg": { "lr": 0.006, "wd": 0.0 }, "mean": 9.905518293380737 }, { "cfg": { "lr": 0.006, "wd": 0.0001 }, "mean": 9.729963302612305 }, { "cfg": { "lr": 0.01, "wd": 0.0 }, "mean": 7.754182696342468 }, { "cfg": { "lr": 0.01, "wd": 0.0001 }, "mean": 7.6093244552612305 } ], "best_cfg": { "lr": 0.01, "wd": 0.0001 }, "full": { "per_seed": [ 7.102456092834473, 8.766936302185059, 7.349897861480713, 7.218007564544678, 6.500193119049072, 7.548990249633789, 8.126578330993652, 8.261975288391113 ], "mean": 7.609379351139069, "std": 0.6841718521926304, "n": 8, "config": { "lr": 0.01, "wd": 0.0001 } }, "audit_runs": [ { "metric": 7.102456092834473, "history": [ 214.53998596191406, 150.79551391601564, 45.06250183105469, 44.90724487304688, 22.644885025024415, 22.263154296875, 24.194852905273436, 15.885811157226563, 17.668506774902344, 14.434557914733887, 13.434552192687988, 12.095326461791991, 11.769148025512695, 10.334355964660645, 10.175729064941406, 8.749533672332763, 8.571477165222168, 7.892980594635009, 7.619370031356811, 7.219865341186523 ] }, { "metric": 8.766936302185059, "history": [ 216.17651611328125, 156.86871520996092, 45.16012939453125, 50.300657958984374, 22.345151824951174, 21.229163131713868, 21.279377746582032, 13.878670806884765, 16.061290435791015, 13.24773666381836, 12.390899543762208, 11.70387607574463, 11.164104766845703, 10.362736930847168, 10.45274757385254, 9.97411792755127, 8.932565460205078, 8.518637466430665, 8.128562622070312, 7.766393203735351 ] }, { "metric": 7.349897861480713, "history": [ 219.39349243164062, 146.39451263427733, 36.564744262695314, 48.9598583984375, 19.486346893310547, 27.214415740966796, 19.930125961303713, 16.0131876373291, 18.18076370239258, 12.916956748962402, 15.148296737670899, 11.739884834289551, 10.988883819580078, 9.74200210571289, 9.59509126663208, 8.717781295776367, 8.149822845458985, 7.8048429107666015, 7.387167625427246, 7.143137340545654 ] }, { "metric": 7.218007564544678, "history": [ 215.29245727539063, 153.3181286621094, 43.44981231689453, 51.55850280761719, 24.195310096740723, 24.33369873046875, 29.513503189086915, 17.208366165161134, 16.743267669677735, 16.64546733856201, 13.176614532470703, 13.314536399841309, 12.080739440917968, 11.744078483581543, 11.132203235626221, 10.857590484619141, 9.981960220336914, 9.732174415588378, 9.127348098754883, 8.670081977844239 ] }, { "metric": 6.500193119049072, "history": [ 232.4084051513672, 140.419755859375, 30.65204315185547, 56.73716537475586, 19.858126831054687, 31.676787109375, 23.139746131896974, 17.12270294189453, 19.785086517333983, 13.713419494628907, 15.035507736206055, 12.384070320129394, 11.58296558380127, 10.596111755371094, 9.830599327087402, 8.959623203277587, 8.42355484008789, 7.857340164184571, 7.7755109977722165, 7.369232406616211 ] }, { "metric": 7.548990249633789, "history": [ 218.01401733398438, 161.49740661621092, 52.60572364807129, 45.408958740234375, 24.290284996032714, 19.144558486938475, 21.734841079711913, 13.96752471923828, 13.811888771057129, 13.178979797363281, 11.018290138244629, 11.17254119873047, 9.871922531127929, 9.451333503723145, 8.699445419311523, 8.35977834701538, 7.6648030090332036, 7.595121612548828, 7.025888919830322, 7.046415634155274 ] }, { "metric": 8.126578330993652, "history": [ 219.12206787109375, 150.0679965209961, 39.91819145202637, 53.258799896240234, 22.7507950592041, 25.290062026977537, 24.593091125488282, 16.55346389770508, 17.977804107666017, 14.593553962707519, 14.882343139648437, 13.11092544555664, 12.004822521209716, 11.090541763305664, 10.585743408203125, 9.785175476074219, 9.138872356414796, 8.626234855651855, 8.139089069366456, 7.8595320510864255 ] }, { "metric": 8.261975288391113, "history": [ 235.50824462890625, 181.1517953491211, 66.92042526245118, 44.197750854492185, 29.396741561889648, 23.55883369445801, 26.864130783081055, 17.323555297851563, 18.29806915283203, 16.233560333251955, 14.512285614013672, 13.535792064666747, 11.563432502746583, 11.270505142211913, 9.929189453125, 9.247753982543946, 8.481544513702392, 7.953108673095703, 7.3793183708190915, 7.1775992965698245 ] } ] }, "idea": { "per_seed": [ 8.10361099243164, 8.1812162399292, 7.210653305053711, 8.95467758178711, 7.580370903015137, 8.632294654846191, 8.847290992736816, 9.434948921203613 ], "mean": 8.368132948875427, "std": 0.6934283174375584, "n": 8, "config": { "lr": 0.01, "r": 0.01, "beta": 1.3 }, "diagnostics": [ { "metric": 8.10361099243164, "restarts": 3, "max_lr": 0.017332530178673953, "update_norm_mean": 0.12184708876397417, "update_growth_mean": 1.0122137089134595 }, { "metric": 8.1812162399292, "restarts": 3, "max_lr": 0.014769807938826427, "update_norm_mean": 0.11243514141108242, "update_growth_mean": 1.0269314387105273 }, { "metric": 7.210653305053711, "restarts": 3, "max_lr": 0.014769807938826427, "update_norm_mean": 0.11399464353001698, "update_growth_mean": 1.0031604459630117 }, { "metric": 8.95467758178711, "restarts": 0, "max_lr": 0.02203396426255937, "update_norm_mean": 0.1521656754655411, "update_growth_mean": 1.0103780184537154 }, { "metric": 7.580370903015137, "restarts": 0, "max_lr": 0.02203396426255937, "update_norm_mean": 0.151965435222015, "update_growth_mean": 1.0102908595171285 }, { "metric": 8.632294654846191, "restarts": 2, "max_lr": 0.016820276496988865, "update_norm_mean": 0.11849648705603952, "update_growth_mean": 1.0099841345527327 }, { "metric": 8.847290992736816, "restarts": 4, "max_lr": 0.01648721270700128, "update_norm_mean": 0.1123172825782556, "update_growth_mean": 1.024348321067094 }, { "metric": 9.434948921203613, "restarts": 4, "max_lr": 0.015219615556186337, "update_norm_mean": 0.11368357661240187, "update_growth_mean": 1.0083068740327106 } ] }, "comparison": { "delta_mean": 0.7587535977363586, "idea_wins": 2, "n_pairs": 8, "per_seed_diffs": [ 1.001154899597168, -0.5857200622558594, -0.13924455642700195, 1.7366700172424316, 1.0801777839660645, 1.0833044052124023, 0.7207126617431641, 1.1729736328125 ], "p_value": 0.04105, "mde": 0.632922491434946, "mde_rel_pct": 8.317662482423119, "verdict": "idea worse (significant)", "system_worked": false }, "search_space_parity": true, "baseline_knobs_swept": [ "lr", "weight_decay" ], "mechanism_signature": { "prediction": "restart criterion detects update growth at least beta*exp(r) schedule growth", "predicted_threshold": 1.3130652172094184, "observed_mean_growth_over_schedule": 1.0031201995404122, "observed_restarts_mean": 2.375, "confirmed": true } }