{ "overall": { "count": 6113, "accuracy": 0.7906101750368068, "brier": 0.2865185962820081, "brier_soft": 0.26187629556812925, "nll": 0.531438177468442, "nll_soft": 0.5442780755003573, "ece": 0.027146574970523422, "mean_confidence": 0.7952999926910861, "confident_error_rate": 0.017176509079011942, "score_mae": 0.5416698742073451, "evidence_auc": 0.9997787857086019 }, "by_type": { "bool": { "count": 1898, "accuracy": 0.8535300316122234, "brier": 0.20239683530024727, "brier_soft": 0.1986991046710315, "nll": 0.3164081539673983, "nll_soft": 0.320539412862896, "ece": 0.03776428708532474, "mean_confidence": 0.891294318697548, "confident_error_rate": 0.02476290832455216, "evidence_auc": 0.9991618000687049 }, "choice": { "count": 3248, "accuracy": 0.7924876847290641, "brier": 0.284974599802582, "brier_soft": 0.26098893765800385, "nll": 0.5679457318609575, "nll_soft": 0.5810034227604031, "ece": 0.030920124238874132, "mean_confidence": 0.8102787767075924, "confident_error_rate": 0.017857142857142856, "evidence_auc": 0.999926113573992 }, "score": { "count": 967, "accuracy": 0.6608066184074457, "brier": 0.45681642762488106, "brier_soft": 0.38885917748620424, "nll": 0.8308698713030757, "nll_soft": 0.860071305888437, "ece": 0.1165943047968045, "mean_confidence": 0.5565737039156183, "confident_error_rate": 0.0, "score_mae": 0.5416698742073451 } }, "by_source": { "ag_news": { "count": 100, "accuracy": 0.91, "brier": 0.14107379849283894, "brier_soft": 0.14107379849283894, "nll": 0.2626451906685023, "nll_soft": 0.2626451906685023, "ece": 0.04884664500994877, "mean_confidence": 0.9267561964206152, "confident_error_rate": 0.03 }, "amazon_polarity": { "count": 100, "accuracy": 0.93, "brier": 0.08932649085635297, "brier_soft": 0.08932649085635297, "nll": 0.1286234055208405, "nll_soft": 0.1286234055208405, "ece": 0.0598796401142421, "mean_confidence": 0.9619265198646406, "confident_error_rate": 0.01 }, "amazon_stars": { "count": 100, "accuracy": 0.66, "brier": 0.5146504076696957, "brier_soft": 0.5146504076696957, "nll": 0.9000029896261006, "nll_soft": 0.9000029896261006, "ece": 0.15304034642163666, "mean_confidence": 0.5267374344918574, "confident_error_rate": 0.0, "score_mae": 0.5121194916103842 }, "anli": { "count": 100, "accuracy": 0.58, "brier": 0.5707019496568548, "brier_soft": 0.5707019496568548, "nll": 0.9733660613496383, "nll_soft": 0.9733660613496383, "ece": 0.20588943546049263, "mean_confidence": 0.781207916786596, "confident_error_rate": 0.06 }, "arc_challenge": { "count": 100, "accuracy": 0.92, "brier": 0.138929125961344, "brier_soft": 0.138929125961344, "nll": 0.24651074556516497, "nll_soft": 0.24651074556516497, "ece": 0.07689403844076734, "mean_confidence": 0.9312481481859443, "confident_error_rate": 0.03 }, "arc_easy": { "count": 100, "accuracy": 0.96, "brier": 0.06105675437719455, "brier_soft": 0.06105675437719455, "nll": 0.1295253280350266, "nll_soft": 0.1295253280350266, "ece": 0.05505374327055725, "mean_confidence": 0.9480651360916019, "confident_error_rate": 0.0 }, "banking77": { "count": 100, "accuracy": 0.73, "brier": 0.3608873627059889, "brier_soft": 0.3608873627059889, "nll": 0.9405229381993333, "nll_soft": 0.9405229381993333, "ece": 0.08553226405095403, "mean_confidence": 0.7356290730340809, "confident_error_rate": 0.0 }, "boolq": { "count": 125, "accuracy": 0.768, "brier": 0.29296620551098923, "brier_soft": 0.29296620551098923, "nll": 0.4262508125125178, "nll_soft": 0.4262508125125178, "ece": 0.06190941058686133, "mean_confidence": 0.7972549868134068, "confident_error_rate": 0.008, "evidence_auc": 1.0 }, "cb": { "count": 28, "accuracy": 0.9285714285714286, "brier": 0.08452813725424528, "brier_soft": 0.08452813725424528, "nll": 0.15482732377842925, "nll_soft": 0.15482732377842925, "ece": 0.0891914767253553, "mean_confidence": 0.9239020966140715, "confident_error_rate": 0.0 }, "civil_comments_severity": { "count": 100, "accuracy": 0.84, "brier": 0.2854281637857832, "brier_soft": 0.18639637414257712, "nll": 0.5742660112377674, "nll_soft": 0.6703868491946997, "ece": 0.17662348482102086, "mean_confidence": 0.683961388574109, "confident_error_rate": 0.0, "score_mae": 0.4394353714638848 }, "civil_comments_toxic": { "count": 100, "accuracy": 0.93, "brier": 0.11580263606349772, "brier_soft": 0.045619708720982234, "nll": 0.2174061558177036, "nll_soft": 0.2958174496542495, "ece": 0.07866423697912733, "mean_confidence": 0.8840980723553749, "confident_error_rate": 0.01 }, "clinc150": { "count": 100, "accuracy": 0.87, "brier": 0.21760990606015762, "brier_soft": 0.21760990606015762, "nll": 0.6578076778338765, "nll_soft": 0.6578076778338765, "ece": 0.0802484366750311, "mean_confidence": 0.8693127305899719, "confident_error_rate": 0.02 }, "cola": { "count": 100, "accuracy": 0.75, "brier": 0.3132800525413747, "brier_soft": 0.3132800525413747, "nll": 0.4747007129405839, "nll_soft": 0.4747007129405839, "ece": 0.11109424747890481, "mean_confidence": 0.8610942474789048, "confident_error_rate": 0.04 }, "commonsense_qa": { "count": 100, "accuracy": 0.87, "brier": 0.19463212383948517, "brier_soft": 0.19463212383948517, "nll": 0.36843752512773237, "nll_soft": 0.36843752512773237, "ece": 0.04958840880497705, "mean_confidence": 0.8938470752943246, "confident_error_rate": 0.02 }, "cosmos_qa": { "count": 120, "accuracy": 0.7333333333333333, "brier": 0.3977299966460981, "brier_soft": 0.3977299966460981, "nll": 0.7969430905955159, "nll_soft": 0.7969430905955159, "ece": 0.11449862878263802, "mean_confidence": 0.7867019719064755, "confident_error_rate": 0.016666666666666666, "evidence_auc": 0.9990625 }, "dates": { "count": 300, "accuracy": 0.9066666666666666, "brier": 0.1366547604463823, "brier_soft": 0.1366547604463823, "nll": 0.22760700905623565, "nll_soft": 0.22760700905623565, "ece": 0.042681842028014755, "mean_confidence": 0.8748727759673026, "confident_error_rate": 0.006666666666666667, "score_mae": 0.2819932523078008 }, "dbpedia": { "count": 100, "accuracy": 1.0, "brier": 0.00018795749398129978, "brier_soft": 0.00018795749398129978, "nll": 0.008357058821629917, "nll_soft": 0.008357058821629917, "ece": 0.00828557394777929, "mean_confidence": 0.9917144260522207, "confident_error_rate": 0.0 }, "dynasent": { "count": 100, "accuracy": 0.69, "brier": 0.4323214023458542, "brier_soft": 0.30704619662116733, "nll": 0.7412403519607587, "nll_soft": 0.790334108879113, "ece": 0.11594177635826394, "mean_confidence": 0.5767413146517898, "confident_error_rate": 0.0, "score_mae": 0.5042056048814512 }, "enron_spam": { "count": 100, "accuracy": 0.95, "brier": 0.061530910842468194, "brier_soft": 0.061530910842468194, "nll": 0.09922311989131462, "nll_soft": 0.09922311989131462, "ece": 0.04473396168787776, "mean_confidence": 0.9767910137716824, "confident_error_rate": 0.02 }, "go_emotions": { "count": 100, "accuracy": 0.54, "brier": 0.6541717455466549, "brier_soft": 0.25009065256646124, "nll": 1.652207905091813, "nll_soft": 2.060946796357424, "ece": 0.18927007018484554, "mean_confidence": 0.35338819562567303, "confident_error_rate": 0.0 }, "hellaswag": { "count": 100, "accuracy": 0.88, "brier": 0.16797113212596798, "brier_soft": 0.16797113212596798, "nll": 0.3132057670542855, "nll_soft": 0.3132057670542855, "ece": 0.07032309961111621, "mean_confidence": 0.9228263155853855, "confident_error_rate": 0.02 }, "imdb": { "count": 100, "accuracy": 0.94, "brier": 0.09338526335431517, "brier_soft": 0.09338526335431517, "nll": 0.15406285631818842, "nll_soft": 0.15406285631818842, "ece": 0.043104711718764975, "mean_confidence": 0.959424845682162, "confident_error_rate": 0.01 }, "ledgar": { "count": 100, "accuracy": 0.75, "brier": 0.3506827260603255, "brier_soft": 0.3506827260603255, "nll": 0.9000206841298396, "nll_soft": 0.9000206841298396, "ece": 0.08742405102661228, "mean_confidence": 0.7466266913748092, "confident_error_rate": 0.01 }, "liar": { "count": 100, "accuracy": 0.24, "brier": 0.7684441156135192, "brier_soft": 0.7684441156135192, "nll": 1.5789539284601737, "nll_soft": 1.5789539284601737, "ece": 0.09676656231782328, "mean_confidence": 0.30260542335327906, "confident_error_rate": 0.0, "score_mae": 1.127111411685787 }, "measuring_hate_speech": { "count": 100, "accuracy": 0.84, "brier": 0.30059922331895095, "brier_soft": 0.15922982151650258, "nll": 0.5642970152096843, "nll_soft": 0.6878727674961508, "ece": 0.196925863260578, "mean_confidence": 0.6452313291559981, "confident_error_rate": 0.0, "score_mae": 0.6025989257232528 }, "medqa": { "count": 100, "accuracy": 0.61, "brier": 0.501600476205162, "brier_soft": 0.501600476205162, "nll": 0.9209206864946542, "nll_soft": 0.9209206864946542, "ece": 0.10855441863453118, "mean_confidence": 0.6787774204404297, "confident_error_rate": 0.01 }, "mnli": { "count": 100, "accuracy": 0.9, "brier": 0.13846939435764263, "brier_soft": 0.13846939435764263, "nll": 0.2625242881689338, "nll_soft": 0.2625242881689338, "ece": 0.05978230532069495, "mean_confidence": 0.867065953881717, "confident_error_rate": 0.0 }, "mrpc": { "count": 100, "accuracy": 0.81, "brier": 0.26125558901505913, "brier_soft": 0.26125558901505913, "nll": 0.3984524563684818, "nll_soft": 0.3984524563684818, "ece": 0.09004270964182014, "mean_confidence": 0.8107605511150068, "confident_error_rate": 0.0 }, "multirc": { "count": 125, "accuracy": 0.752, "brier": 0.3193987297602871, "brier_soft": 0.3193987297602871, "nll": 0.47067874987325165, "nll_soft": 0.47067874987325165, "ece": 0.0981009103037617, "mean_confidence": 0.79820435302511, "confident_error_rate": 0.016, "evidence_auc": 0.9989333333333333 }, "openbookqa": { "count": 100, "accuracy": 0.87, "brier": 0.1793119275121386, "brier_soft": 0.1793119275121386, "nll": 0.3212052656163398, "nll_soft": 0.3212052656163398, "ece": 0.048824714093196483, "mean_confidence": 0.8795181984743128, "confident_error_rate": 0.0 }, "piqa": { "count": 100, "accuracy": 0.82, "brier": 0.24895208825020151, "brier_soft": 0.24895208825020151, "nll": 0.389222560310709, "nll_soft": 0.389222560310709, "ece": 0.08642289363272926, "mean_confidence": 0.838094799621635, "confident_error_rate": 0.03 }, "policy_application": { "count": 150, "accuracy": 0.9066666666666666, "brier": 0.139494695459916, "brier_soft": 0.139494695459916, "nll": 0.21704375203377746, "nll_soft": 0.21704375203377746, "ece": 0.059896815803595904, "mean_confidence": 0.9398540889590621, "confident_error_rate": 0.013333333333333334, "evidence_auc": 1.0 }, "policy_hard": { "count": 300, "accuracy": 0.8433333333333334, "brier": 0.20894731883691625, "brier_soft": 0.20894731883691625, "nll": 0.34797745624001836, "nll_soft": 0.34797745624001836, "ece": 0.040093495814151, "mean_confidence": 0.880565356561097, "confident_error_rate": 0.02666666666666667, "evidence_auc": 0.9817708333333334 }, "qqp": { "count": 100, "accuracy": 0.91, "brier": 0.13989046875222017, "brier_soft": 0.13989046875222017, "nll": 0.234108067003616, "nll_soft": 0.234108067003616, "ece": 0.05421047729337747, "mean_confidence": 0.8683763009029548, "confident_error_rate": 0.0 }, "race": { "count": 125, "accuracy": 0.72, "brier": 0.3468329954783986, "brier_soft": 0.3468329954783986, "nll": 0.6595321747170682, "nll_soft": 0.6595321747170682, "ece": 0.08253622206480249, "mean_confidence": 0.7759816152635733, "confident_error_rate": 0.024, "evidence_auc": 0.9992 }, "rotten_tomatoes": { "count": 100, "accuracy": 0.94, "brier": 0.07896598050795499, "brier_soft": 0.07896598050795499, "nll": 0.1295703707449942, "nll_soft": 0.1295703707449942, "ece": 0.06612016556692232, "mean_confidence": 0.9349814985533582, "confident_error_rate": 0.0 }, "rte": { "count": 100, "accuracy": 0.91, "brier": 0.1785855924400104, "brier_soft": 0.1785855924400104, "nll": 0.3144416684299057, "nll_soft": 0.3144416684299057, "ece": 0.08369600771650051, "mean_confidence": 0.9264092676999237, "confident_error_rate": 0.05 }, "scotus": { "count": 100, "accuracy": 0.21, "brier": 0.8933246384582364, "brier_soft": 0.8933246384582364, "nll": 2.4025798961358653, "nll_soft": 2.4025798961358653, "ece": 0.026259098193438857, "mean_confidence": 0.20252531874914392, "confident_error_rate": 0.0 }, "shp": { "count": 100, "accuracy": 0.68, "brier": 0.47749577158017964, "brier_soft": 0.1025225581044758, "nll": 0.6706076521115167, "nll_soft": 0.6859825612598958, "ece": 0.1478788162760493, "mean_confidence": 0.5369806117053606, "confident_error_rate": 0.0 }, "sms_spam": { "count": 100, "accuracy": 0.99, "brier": 0.018716052570869043, "brier_soft": 0.018716052570869043, "nll": 0.04010322700735336, "nll_soft": 0.04010322700735336, "ece": 0.002744809770715926, "mean_confidence": 0.9907091112815634, "confident_error_rate": 0.01 }, "snli": { "count": 100, "accuracy": 0.88, "brier": 0.16836393178606746, "brier_soft": 0.16836393178606746, "nll": 0.2924678063682347, "nll_soft": 0.2924678063682347, "ece": 0.034565203155391265, "mean_confidence": 0.8986885605645243, "confident_error_rate": 0.03 }, "social_iqa": { "count": 100, "accuracy": 0.78, "brier": 0.30628460928922563, "brier_soft": 0.30628460928922563, "nll": 0.5561253043672776, "nll_soft": 0.5561253043672776, "ece": 0.10795976232448823, "mean_confidence": 0.8395705446442819, "confident_error_rate": 0.03 }, "sst2": { "count": 100, "accuracy": 0.96, "brier": 0.07928032002666491, "brier_soft": 0.07928032002666491, "nll": 0.15856545014783774, "nll_soft": 0.15856545014783774, "ece": 0.03715473937547079, "mean_confidence": 0.959105934637087, "confident_error_rate": 0.02 }, "sst5": { "count": 100, "accuracy": 0.67, "brier": 0.5047456075070107, "brier_soft": 0.5047456075070107, "nll": 0.9046301136653718, "nll_soft": 0.9046301136653718, "ece": 0.1989545581620594, "mean_confidence": 0.4993328109434478, "confident_error_rate": 0.0, "score_mae": 0.46473458149541735 }, "stance_abortion": { "count": 60, "accuracy": 0.6666666666666666, "brier": 0.4318499648737531, "brier_soft": 0.4318499648737531, "nll": 0.6693344974312939, "nll_soft": 0.6693344974312939, "ece": 0.16791879623099454, "mean_confidence": 0.7610590780360661, "confident_error_rate": 0.016666666666666666 }, "stance_atheism": { "count": 60, "accuracy": 0.7833333333333333, "brier": 0.3150497260676467, "brier_soft": 0.3150497260676467, "nll": 0.5028852417656949, "nll_soft": 0.5028852417656949, "ece": 0.1430124566231118, "mean_confidence": 0.8368338633078285, "confident_error_rate": 0.016666666666666666 }, "stance_feminist": { "count": 60, "accuracy": 0.75, "brier": 0.3648047830860601, "brier_soft": 0.3648047830860601, "nll": 0.5867590544526551, "nll_soft": 0.5867590544526551, "ece": 0.13802642902758205, "mean_confidence": 0.779361470377674, "confident_error_rate": 0.0 }, "stance_hillary": { "count": 60, "accuracy": 0.6666666666666666, "brier": 0.39158167992187276, "brier_soft": 0.39158167992187276, "nll": 0.6568302459427436, "nll_soft": 0.6568302459427436, "ece": 0.13117570539181508, "mean_confidence": 0.7419570046865254, "confident_error_rate": 0.0 }, "stsb": { "count": 100, "accuracy": 0.54, "brier": 0.5570610302874376, "brier_soft": 0.2655908186167737, "nll": 0.9685020049260296, "nll_soft": 0.9820895302047219, "ece": 0.07001749408544906, "mean_confidence": 0.539751064816704, "confident_error_rate": 0.0, "score_mae": 0.565603055088638 }, "subjectivity": { "count": 100, "accuracy": 0.94, "brier": 0.09908461321243932, "brier_soft": 0.09908461321243932, "nll": 0.16222277170135949, "nll_soft": 0.16222277170135949, "ece": 0.04927684333663743, "mean_confidence": 0.9490918012704767, "confident_error_rate": 0.01 }, "swag": { "count": 100, "accuracy": 0.77, "brier": 0.3000022704804431, "brier_soft": 0.3000022704804431, "nll": 0.5425527902770044, "nll_soft": 0.5425527902770044, "ece": 0.07660043607599595, "mean_confidence": 0.7782203540696726, "confident_error_rate": 0.01 }, "trec": { "count": 100, "accuracy": 0.97, "brier": 0.061975700748181185, "brier_soft": 0.061975700748181185, "nll": 0.15411059891893825, "nll_soft": 0.15411059891893825, "ece": 0.017816031807190158, "mean_confidence": 0.9619683086236548, "confident_error_rate": 0.02 }, "tweet_hate": { "count": 100, "accuracy": 0.48, "brier": 0.7473889756686806, "brier_soft": 0.7473889756686806, "nll": 1.1776300863149485, "nll_soft": 1.1776300863149485, "ece": 0.36222919958424477, "mean_confidence": 0.8422291995842447, "confident_error_rate": 0.21 }, "tweet_sentiment": { "count": 100, "accuracy": 0.66, "brier": 0.4403312511848521, "brier_soft": 0.4403312511848521, "nll": 0.7376234445010855, "nll_soft": 0.7376234445010855, "ece": 0.10092449734474333, "mean_confidence": 0.5835192824734373, "confident_error_rate": 0.0, "score_mae": 0.3924732269672393 }, "vitaminc": { "count": 100, "accuracy": 0.84, "brier": 0.26744036733920845, "brier_soft": 0.26744036733920845, "nll": 0.5030591509155484, "nll_soft": 0.5030591509155484, "ece": 0.08253130737011233, "mean_confidence": 0.8672411180861649, "confident_error_rate": 0.05 }, "wic": { "count": 100, "accuracy": 0.71, "brier": 0.35517493029561253, "brier_soft": 0.35517493029561253, "nll": 0.5259271797133195, "nll_soft": 0.5259271797133195, "ece": 0.06847744577518106, "mean_confidence": 0.748921172428449, "confident_error_rate": 0.0 }, "winogrande": { "count": 100, "accuracy": 0.75, "brier": 0.37336827230181, "brier_soft": 0.37336827230181, "nll": 0.6051925483123668, "nll_soft": 0.6051925483123668, "ece": 0.1422213440002727, "mean_confidence": 0.8710624725593938, "confident_error_rate": 0.07 }, "yelp": { "count": 100, "accuracy": 0.68, "brier": 0.4571774209762717, "brier_soft": 0.4571774209762717, "nll": 0.7820219928182245, "nll_soft": 0.7820219928182245, "ece": 0.1450263335660819, "mean_confidence": 0.5498948735721532, "confident_error_rate": 0.0, "score_mae": 0.44073053562274644 } }, "reliability": [ { "bin": "0.0-0.1", "count": 0, "confidence": NaN, "accuracy": NaN }, { "bin": "0.1-0.2", "count": 74, "confidence": 0.17221642633416331, "accuracy": 0.1891891891891892 }, { "bin": "0.2-0.3", "count": 143, "confidence": 0.2449173694172205, "accuracy": 0.25874125874125875 }, { "bin": "0.3-0.4", "count": 158, "confidence": 0.35688169328104796, "accuracy": 0.34177215189873417 }, { "bin": "0.4-0.5", "count": 437, "confidence": 0.46054809633778104, "accuracy": 0.5560640732265446 }, { "bin": "0.5-0.6", "count": 621, "confidence": 0.5505989146559541, "accuracy": 0.5861513687600645 }, { "bin": "0.6-0.7", "count": 475, "confidence": 0.6529554400217262, "accuracy": 0.6589473684210526 }, { "bin": "0.7-0.8", "count": 485, "confidence": 0.752970305163874, "accuracy": 0.709278350515464 }, { "bin": "0.8-0.9", "count": 677, "confidence": 0.8544566232414949, "accuracy": 0.7769571639586411 }, { "bin": "0.9-1.0", "count": 3043, "confidence": 0.9728956090557227, "accuracy": 0.9654945777193559 } ], "seconds": 67.44323062896729, "model": "runs/assay-4b-v4", "data": "data/v4/dev.jsonl" }