{ "overall": { "count": 6113, "accuracy": 0.8081138557173237, "brier": 0.26832227363299765, "brier_soft": 0.24630130870612255, "nll": 0.505385205242332, "nll_soft": 0.523684859731946, "ece": 0.03634856478866526, "mean_confidence": 0.8375686736383623, "confident_error_rate": 0.03140847374447898, "score_mae": 0.485172839677863, "evidence_auc": 0.9938195273334924 }, "by_type": { "bool": { "count": 1898, "accuracy": 0.8730242360379347, "brier": 0.18502372315079466, "brier_soft": 0.18166443929411372, "nll": 0.293633727648158, "nll_soft": 0.2986379499634176, "ece": 0.04287967945534256, "mean_confidence": 0.9132292502672847, "confident_error_rate": 0.033192834562697573, "evidence_auc": 0.993037902210008 }, "choice": { "count": 3248, "accuracy": 0.8075738916256158, "brier": 0.27162731616245755, "brier_soft": 0.2490854051600207, "nll": 0.5523529039003323, "nll_soft": 0.5704940630171196, "ece": 0.05672168252811685, "mean_confidence": 0.8598830180546172, "confident_error_rate": 0.03940886699507389, "evidence_auc": 0.9943635212159595 }, "score": { "count": 967, "accuracy": 0.6825232678386763, "brier": 0.42071717609373765, "brier_soft": 0.3638173716448313, "nll": 0.763247893176724, "nll_soft": 0.8081758033414839, "ece": 0.07701368480022007, "mean_confidence": 0.6141139010368198, "confident_error_rate": 0.001034126163391934, "score_mae": 0.485172839677863 } }, "by_source": { "ag_news": { "count": 100, "accuracy": 0.89, "brier": 0.17744726851892056, "brier_soft": 0.17744726851892056, "nll": 0.33952497205533755, "nll_soft": 0.33952497205533755, "ece": 0.08231704690785716, "mean_confidence": 0.9545400192391296, "confident_error_rate": 0.05 }, "amazon_polarity": { "count": 100, "accuracy": 0.97, "brier": 0.06681896606224931, "brier_soft": 0.06681896606224931, "nll": 0.11474157606024799, "nll_soft": 0.11474157606024799, "ece": 0.04592720580103715, "mean_confidence": 0.9684370974150603, "confident_error_rate": 0.01 }, "amazon_stars": { "count": 100, "accuracy": 0.58, "brier": 0.513726379766378, "brier_soft": 0.513726379766378, "nll": 0.8877784154246966, "nll_soft": 0.8877784154246966, "ece": 0.07482866258617463, "mean_confidence": 0.5826355888936225, "confident_error_rate": 0.01, "score_mae": 0.5120021688214367 }, "anli": { "count": 100, "accuracy": 0.63, "brier": 0.5701719988732347, "brier_soft": 0.5701719988732347, "nll": 1.0697121018694993, "nll_soft": 1.0697121018694993, "ece": 0.21497286575145572, "mean_confidence": 0.8324542993352486, "confident_error_rate": 0.13 }, "arc_challenge": { "count": 100, "accuracy": 0.94, "brier": 0.10599146992155316, "brier_soft": 0.10599146992155316, "nll": 0.19019848146359147, "nll_soft": 0.19019848146359147, "ece": 0.060081904951066994, "mean_confidence": 0.9680451934575867, "confident_error_rate": 0.02 }, "arc_easy": { "count": 100, "accuracy": 0.97, "brier": 0.04437057638950536, "brier_soft": 0.04437057638950536, "nll": 0.10036992791451768, "nll_soft": 0.10036992791451768, "ece": 0.0319046333779032, "mean_confidence": 0.9737156977945386, "confident_error_rate": 0.0 }, "banking77": { "count": 100, "accuracy": 0.75, "brier": 0.3398126842344269, "brier_soft": 0.3398126842344269, "nll": 0.8138418054666235, "nll_soft": 0.8138418054666235, "ece": 0.10284948389955295, "mean_confidence": 0.8351027846874414, "confident_error_rate": 0.02 }, "boolq": { "count": 125, "accuracy": 0.824, "brier": 0.2450309380139414, "brier_soft": 0.2450309380139414, "nll": 0.3832584543059262, "nll_soft": 0.3832584543059262, "ece": 0.08305327395905993, "mean_confidence": 0.8318427620613265, "confident_error_rate": 0.024, "evidence_auc": 1.0 }, "cb": { "count": 28, "accuracy": 1.0, "brier": 0.012525034629445196, "brier_soft": 0.012525034629445196, "nll": 0.052362560175484685, "nll_soft": 0.052362560175484685, "ece": 0.04811603778085669, "mean_confidence": 0.9518839622191433, "confident_error_rate": 0.0 }, "civil_comments_severity": { "count": 100, "accuracy": 0.86, "brier": 0.23187375352715037, "brier_soft": 0.17771031506273136, "nll": 0.47878826513277256, "nll_soft": 0.6466024107213819, "ece": 0.1323208998113436, "mean_confidence": 0.7884055630186311, "confident_error_rate": 0.0, "score_mae": 0.32065261707897563 }, "civil_comments_toxic": { "count": 100, "accuracy": 0.94, "brier": 0.10944319163588359, "brier_soft": 0.04568398403607942, "nll": 0.19590720265259998, "nll_soft": 0.29088734219622764, "ece": 0.06555426270198156, "mean_confidence": 0.9010713536348429, "confident_error_rate": 0.02 }, "clinc150": { "count": 100, "accuracy": 0.86, "brier": 0.22210552087365046, "brier_soft": 0.22210552087365046, "nll": 0.6395453706356091, "nll_soft": 0.6395453706356091, "ece": 0.11099497317528545, "mean_confidence": 0.9251752116271217, "confident_error_rate": 0.07 }, "cola": { "count": 100, "accuracy": 0.82, "brier": 0.259117219091911, "brier_soft": 0.259117219091911, "nll": 0.40935600764726915, "nll_soft": 0.40935600764726915, "ece": 0.08932037436110461, "mean_confidence": 0.8750779518376584, "confident_error_rate": 0.06 }, "commonsense_qa": { "count": 100, "accuracy": 0.85, "brier": 0.21088569401090138, "brier_soft": 0.21088569401090138, "nll": 0.3965346936975742, "nll_soft": 0.3965346936975742, "ece": 0.07310374823833152, "mean_confidence": 0.9136859527085365, "confident_error_rate": 0.05 }, "cosmos_qa": { "count": 120, "accuracy": 0.75, "brier": 0.3841359793377875, "brier_soft": 0.3841359793377875, "nll": 0.7882912921850803, "nll_soft": 0.7882912921850803, "ece": 0.14439795786405063, "mean_confidence": 0.8702594151182701, "confident_error_rate": 0.08333333333333333, "evidence_auc": 0.999375 }, "dates": { "count": 300, "accuracy": 0.9133333333333333, "brier": 0.12682090218682965, "brier_soft": 0.12682090218682965, "nll": 0.20972857100463424, "nll_soft": 0.20972857100463424, "ece": 0.022666054970448136, "mean_confidence": 0.9010661321285901, "confident_error_rate": 0.016666666666666666, "score_mae": 0.22207703752612915 }, "dbpedia": { "count": 100, "accuracy": 0.99, "brier": 0.005118529799186151, "brier_soft": 0.005118529799186151, "nll": 0.009824460705460551, "nll_soft": 0.009824460705460551, "ece": 0.007555830713154134, "mean_confidence": 0.9922426538540887, "confident_error_rate": 0.0 }, "dynasent": { "count": 100, "accuracy": 0.72, "brier": 0.3964167951058033, "brier_soft": 0.27899306375845145, "nll": 0.6748527018027392, "nll_soft": 0.738852701627455, "ece": 0.13431695655660056, "mean_confidence": 0.6383355955935446, "confident_error_rate": 0.0, "score_mae": 0.45075000945645066 }, "enron_spam": { "count": 100, "accuracy": 0.98, "brier": 0.04102532628851713, "brier_soft": 0.04102532628851713, "nll": 0.0898814377234004, "nll_soft": 0.0898814377234004, "ece": 0.0173370851081312, "mean_confidence": 0.9860527259793832, "confident_error_rate": 0.02 }, "go_emotions": { "count": 100, "accuracy": 0.59, "brier": 0.6037759399411283, "brier_soft": 0.24043395414551444, "nll": 1.4698625936544292, "nll_soft": 2.0367941435838968, "ece": 0.19190927347525633, "mean_confidence": 0.4642116066087375, "confident_error_rate": 0.0 }, "hellaswag": { "count": 100, "accuracy": 0.93, "brier": 0.11090716468636642, "brier_soft": 0.11090716468636642, "nll": 0.2050347391855065, "nll_soft": 0.2050347391855065, "ece": 0.05729168206322114, "mean_confidence": 0.9446148727839713, "confident_error_rate": 0.01 }, "imdb": { "count": 100, "accuracy": 0.94, "brier": 0.07627635078317603, "brier_soft": 0.07627635078317603, "nll": 0.14854959274335647, "nll_soft": 0.14854959274335647, "ece": 0.04784298973029209, "mean_confidence": 0.9738336965806963, "confident_error_rate": 0.02 }, "ledgar": { "count": 100, "accuracy": 0.78, "brier": 0.3408363523310183, "brier_soft": 0.3408363523310183, "nll": 0.7893043506756048, "nll_soft": 0.7893043506756048, "ece": 0.11682861577561969, "mean_confidence": 0.8561558831040632, "confident_error_rate": 0.05 }, "liar": { "count": 100, "accuracy": 0.32, "brier": 0.7419574626836368, "brier_soft": 0.7419574626836368, "nll": 1.491062320585033, "nll_soft": 1.491062320585033, "ece": 0.0781747414321995, "mean_confidence": 0.32746267464577467, "confident_error_rate": 0.0, "score_mae": 1.0138855311286923 }, "measuring_hate_speech": { "count": 100, "accuracy": 0.86, "brier": 0.2815976144439547, "brier_soft": 0.1480667816419444, "nll": 0.5342900214990065, "nll_soft": 0.6664983532839511, "ece": 0.17104543438427827, "mean_confidence": 0.6889545656157217, "confident_error_rate": 0.0, "score_mae": 0.5597911792807024 }, "medqa": { "count": 100, "accuracy": 0.65, "brier": 0.4757799375152343, "brier_soft": 0.4757799375152343, "nll": 0.9227869245729243, "nll_soft": 0.9227869245729243, "ece": 0.12644466791999856, "mean_confidence": 0.76931472502784, "confident_error_rate": 0.05 }, "mnli": { "count": 100, "accuracy": 0.93, "brier": 0.0996316035037227, "brier_soft": 0.0996316035037227, "nll": 0.19154369357221818, "nll_soft": 0.19154369357221818, "ece": 0.04375298064064905, "mean_confidence": 0.9138691136500888, "confident_error_rate": 0.01 }, "mrpc": { "count": 100, "accuracy": 0.81, "brier": 0.24199403452777526, "brier_soft": 0.24199403452777526, "nll": 0.3648608292269973, "nll_soft": 0.3648608292269973, "ece": 0.07234249682670134, "mean_confidence": 0.8719535628687496, "confident_error_rate": 0.02 }, "multirc": { "count": 125, "accuracy": 0.712, "brier": 0.3503515136663073, "brier_soft": 0.3503515136663073, "nll": 0.49143203629023763, "nll_soft": 0.49143203629023763, "ece": 0.14635532511730132, "mean_confidence": 0.8244143309088223, "confident_error_rate": 0.016, "evidence_auc": 0.9816 }, "openbookqa": { "count": 100, "accuracy": 0.89, "brier": 0.17270474379906137, "brier_soft": 0.17270474379906137, "nll": 0.31487342081140474, "nll_soft": 0.31487342081140474, "ece": 0.0706087790568628, "mean_confidence": 0.9002089284314065, "confident_error_rate": 0.02 }, "piqa": { "count": 100, "accuracy": 0.84, "brier": 0.22901713881739658, "brier_soft": 0.22901713881739658, "nll": 0.3914267480993911, "nll_soft": 0.3914267480993911, "ece": 0.09250240078656738, "mean_confidence": 0.8972098554800523, "confident_error_rate": 0.05 }, "policy_application": { "count": 150, "accuracy": 0.9066666666666666, "brier": 0.1202855020589476, "brier_soft": 0.1202855020589476, "nll": 0.17614015484592835, "nll_soft": 0.17614015484592835, "ece": 0.06743290575220404, "mean_confidence": 0.9566756649191206, "confident_error_rate": 0.013333333333333334, "evidence_auc": 1.0 }, "policy_hard": { "count": 300, "accuracy": 0.87, "brier": 0.20656161280894106, "brier_soft": 0.20656161280894106, "nll": 0.35075682790318247, "nll_soft": 0.35075682790318247, "ece": 0.07365686249726952, "mean_confidence": 0.9254342343056455, "confident_error_rate": 0.05333333333333334, "evidence_auc": 0.9759837962962963 }, "qqp": { "count": 100, "accuracy": 0.91, "brier": 0.14255690249181868, "brier_soft": 0.14255690249181868, "nll": 0.22598062142518732, "nll_soft": 0.22598062142518732, "ece": 0.05481505613728288, "mean_confidence": 0.8876121401144598, "confident_error_rate": 0.0 }, "race": { "count": 125, "accuracy": 0.712, "brier": 0.3546762617536465, "brier_soft": 0.3546762617536465, "nll": 0.6863510995706414, "nll_soft": 0.6863510995706414, "ece": 0.1374830827190989, "mean_confidence": 0.8452280203211845, "confident_error_rate": 0.032, "evidence_auc": 0.9816 }, "rotten_tomatoes": { "count": 100, "accuracy": 0.95, "brier": 0.07396981420351864, "brier_soft": 0.07396981420351864, "nll": 0.11963696118574361, "nll_soft": 0.11963696118574361, "ece": 0.04639630588091059, "mean_confidence": 0.9517830907375209, "confident_error_rate": 0.01 }, "rte": { "count": 100, "accuracy": 0.91, "brier": 0.1434888267368668, "brier_soft": 0.1434888267368668, "nll": 0.2490209554286854, "nll_soft": 0.2490209554286854, "ece": 0.058570730504756724, "mean_confidence": 0.9568197067803338, "confident_error_rate": 0.05 }, "scotus": { "count": 100, "accuracy": 0.2, "brier": 0.8886954446327406, "brier_soft": 0.8886954446327406, "nll": 2.419535738172022, "nll_soft": 2.419535738172022, "ece": 0.07677489821988327, "mean_confidence": 0.26455679810075017, "confident_error_rate": 0.0 }, "shp": { "count": 100, "accuracy": 0.59, "brier": 0.4767881507616933, "brier_soft": 0.10796886719815929, "nll": 0.6694358409882295, "nll_soft": 0.6917291391720062, "ece": 0.06946449609206626, "mean_confidence": 0.5484305322637995, "confident_error_rate": 0.0 }, "sms_spam": { "count": 100, "accuracy": 0.99, "brier": 0.01965210244272633, "brier_soft": 0.01965210244272633, "nll": 0.04081128588729513, "nll_soft": 0.04081128588729513, "ece": 0.006474000923922583, "mean_confidence": 0.9920199980802447, "confident_error_rate": 0.01 }, "snli": { "count": 100, "accuracy": 0.95, "brier": 0.08146867312596515, "brier_soft": 0.08146867312596515, "nll": 0.158465964646708, "nll_soft": 0.158465964646708, "ece": 0.028524302433964217, "mean_confidence": 0.9484107171835535, "confident_error_rate": 0.01 }, "social_iqa": { "count": 100, "accuracy": 0.84, "brier": 0.2614318560423284, "brier_soft": 0.2614318560423284, "nll": 0.5328960290627591, "nll_soft": 0.5328960290627591, "ece": 0.08443231200903581, "mean_confidence": 0.8928271961877579, "confident_error_rate": 0.06 }, "sst2": { "count": 100, "accuracy": 0.97, "brier": 0.05705390850807199, "brier_soft": 0.05705390850807199, "nll": 0.12873286798535566, "nll_soft": 0.12873286798535566, "ece": 0.03444267451074121, "mean_confidence": 0.973252100448412, "confident_error_rate": 0.02 }, "sst5": { "count": 100, "accuracy": 0.67, "brier": 0.4731484379764573, "brier_soft": 0.4731484379764573, "nll": 0.820745620437561, "nll_soft": 0.820745620437561, "ece": 0.133823772664781, "mean_confidence": 0.5465827882093774, "confident_error_rate": 0.0, "score_mae": 0.42073229181882404 }, "stance_abortion": { "count": 60, "accuracy": 0.6333333333333333, "brier": 0.5461372518947913, "brier_soft": 0.5461372518947913, "nll": 0.9076919478704142, "nll_soft": 0.9076919478704142, "ece": 0.27791953126303576, "mean_confidence": 0.8731201720022546, "confident_error_rate": 0.13333333333333333 }, "stance_atheism": { "count": 60, "accuracy": 0.7666666666666667, "brier": 0.3807376010727825, "brier_soft": 0.3807376010727825, "nll": 0.6894866431293326, "nll_soft": 0.6894866431293326, "ece": 0.1623851166899183, "mean_confidence": 0.8907779015958398, "confident_error_rate": 0.11666666666666667 }, "stance_feminist": { "count": 60, "accuracy": 0.7166666666666667, "brier": 0.3819212632872141, "brier_soft": 0.3819212632872141, "nll": 0.6580171893898042, "nll_soft": 0.6580171893898042, "ece": 0.13298127336044635, "mean_confidence": 0.825324515184345, "confident_error_rate": 0.05 }, "stance_hillary": { "count": 60, "accuracy": 0.6833333333333333, "brier": 0.45702756655176746, "brier_soft": 0.45702756655176746, "nll": 0.7477193618494712, "nll_soft": 0.7477193618494712, "ece": 0.1903802570948809, "mean_confidence": 0.8185082158591793, "confident_error_rate": 0.03333333333333333 }, "stsb": { "count": 100, "accuracy": 0.67, "brier": 0.4637369660396809, "brier_soft": 0.21863385963253795, "nll": 0.8565117839821994, "nll_soft": 0.9269421980771559, "ece": 0.10975654106260692, "mean_confidence": 0.5983604068227878, "confident_error_rate": 0.0, "score_mae": 0.4717016493911864 }, "subjectivity": { "count": 100, "accuracy": 0.95, "brier": 0.06213683667359613, "brier_soft": 0.06213683667359613, "nll": 0.09836556526336807, "nll_soft": 0.09836556526336807, "ece": 0.037214204902439964, "mean_confidence": 0.965239654729141, "confident_error_rate": 0.01 }, "swag": { "count": 100, "accuracy": 0.82, "brier": 0.261763729502724, "brier_soft": 0.261763729502724, "nll": 0.5414241498614379, "nll_soft": 0.5414241498614379, "ece": 0.07772512379591902, "mean_confidence": 0.8475851911820763, "confident_error_rate": 0.05 }, "trec": { "count": 100, "accuracy": 0.97, "brier": 0.06072142112442041, "brier_soft": 0.06072142112442041, "nll": 0.1440798856068137, "nll_soft": 0.1440798856068137, "ece": 0.02690207486651098, "mean_confidence": 0.9736191389408069, "confident_error_rate": 0.03 }, "tweet_hate": { "count": 100, "accuracy": 0.59, "brier": 0.6465587241335488, "brier_soft": 0.6465587241335488, "nll": 1.033942009148757, "nll_soft": 1.033942009148757, "ece": 0.26985585995709155, "mean_confidence": 0.8472034199650343, "confident_error_rate": 0.2 }, "tweet_sentiment": { "count": 100, "accuracy": 0.64, "brier": 0.4223186571536932, "brier_soft": 0.4223186571536932, "nll": 0.7004327844815618, "nll_soft": 0.7004327844815618, "ece": 0.07372919978481107, "mean_confidence": 0.6227698543011895, "confident_error_rate": 0.0, "score_mae": 0.3842789231307415 }, "vitaminc": { "count": 100, "accuracy": 0.82, "brier": 0.29574730745541433, "brier_soft": 0.29574730745541433, "nll": 0.5826839900533985, "nll_soft": 0.5826839900533985, "ece": 0.11659560936407634, "mean_confidence": 0.9001589414913586, "confident_error_rate": 0.06 }, "wic": { "count": 100, "accuracy": 0.76, "brier": 0.33426080682454734, "brier_soft": 0.33426080682454734, "nll": 0.5083526804189769, "nll_soft": 0.5083526804189769, "ece": 0.06690578964005471, "mean_confidence": 0.8040500963252548, "confident_error_rate": 0.03 }, "winogrande": { "count": 100, "accuracy": 0.84, "brier": 0.25538046579627915, "brier_soft": 0.25538046579627915, "nll": 0.45847041026936064, "nll_soft": 0.45847041026936064, "ece": 0.08929165536123829, "mean_confidence": 0.9104076795837307, "confident_error_rate": 0.07 }, "yelp": { "count": 100, "accuracy": 0.69, "brier": 0.4207442539736879, "brier_soft": 0.4207442539736879, "nll": 0.7108912277650302, "nll_soft": 0.7108912277650302, "ece": 0.09729612455914759, "mean_confidence": 0.629268115548149, "confident_error_rate": 0.0, "score_mae": 0.4090353744354196 } }, "reliability": [ { "bin": "0.0-0.1", "count": 0, "confidence": NaN, "accuracy": NaN }, { "bin": "0.1-0.2", "count": 5, "confidence": 0.19399414131767895, "accuracy": 0.2 }, { "bin": "0.2-0.3", "count": 150, "confidence": 0.2521147949419595, "accuracy": 0.22666666666666666 }, { "bin": "0.3-0.4", "count": 99, "confidence": 0.3545488043431464, "accuracy": 0.3939393939393939 }, { "bin": "0.4-0.5", "count": 348, "confidence": 0.464996527886961, "accuracy": 0.4942528735632184 }, { "bin": "0.5-0.6", "count": 484, "confidence": 0.5522991218018702, "accuracy": 0.5516528925619835 }, { "bin": "0.6-0.7", "count": 425, "confidence": 0.6444857209509922, "accuracy": 0.6494117647058824 }, { "bin": "0.7-0.8", "count": 432, "confidence": 0.748695643337274, "accuracy": 0.7013888888888888 }, { "bin": "0.8-0.9", "count": 547, "confidence": 0.8524781590758739, "accuracy": 0.7623400365630713 }, { "bin": "0.9-1.0", "count": 3623, "confidence": 0.980786561241046, "accuracy": 0.9470052442727022 } ], "seconds": 110.56856632232666, "model": "runs/assay-8b", "data": "data/v4/dev.jsonl" }