{ "overall": { "count": 6113, "accuracy": 0.8081138557173237, "brier": 0.2661494358472481, "brier_soft": 0.24260286062875672, "nll": 0.49575501723827897, "nll_soft": 0.5104445448249281, "ece": 0.024109620444869757, "mean_confidence": 0.8061256466219281, "confident_error_rate": 0.019466710289546867, "score_mae": 0.5130282987896484, "evidence_auc": 0.9938195273334924 }, "by_type": { "bool": { "count": 1898, "accuracy": 0.8730242360379347, "brier": 0.17989218003516222, "brier_soft": 0.17618496083298144, "nll": 0.28048398585407075, "nll_soft": 0.28450098328570933, "ece": 0.021932010938761043, "mean_confidence": 0.8927613096867237, "confident_error_rate": 0.02107481559536354, "evidence_auc": 0.993037902210008 }, "choice": { "count": 3248, "accuracy": 0.8075738916256158, "brier": 0.265855969310664, "brier_soft": 0.24213587499460246, "nll": 0.531371360305088, "nll_soft": 0.5459336605013914, "ece": 0.031153686026911636, "mean_confidence": 0.8267506704226892, "confident_error_rate": 0.02432266009852217, "evidence_auc": 0.9943635212159595 }, "score": { "count": 967, "accuracy": 0.6825232678386763, "brier": 0.4364384232745116, "brier_soft": 0.3745345495140871, "nll": 0.7986531923016, "nll_soft": 0.834717794136494, "ece": 0.12188649461723472, "mean_confidence": 0.5668034482746123, "confident_error_rate": 0.0, "score_mae": 0.5130282987896484 } }, "by_source": { "ag_news": { "count": 100, "accuracy": 0.89, "brier": 0.16845385243161523, "brier_soft": 0.16845385243161523, "nll": 0.31026867024327776, "nll_soft": 0.31026867024327776, "ece": 0.07299394846970284, "mean_confidence": 0.9300757047424044, "confident_error_rate": 0.04 }, "amazon_polarity": { "count": 100, "accuracy": 0.97, "brier": 0.06600077240643171, "brier_soft": 0.06600077240643171, "nll": 0.11152031928475888, "nll_soft": 0.11152031928475888, "ece": 0.045533713599151, "mean_confidence": 0.957777396191883, "confident_error_rate": 0.01 }, "amazon_stars": { "count": 100, "accuracy": 0.58, "brier": 0.5194019417386688, "brier_soft": 0.5194019417386688, "nll": 0.9101515599872236, "nll_soft": 0.9101515599872236, "ece": 0.06545616524854728, "mean_confidence": 0.5352706927405049, "confident_error_rate": 0.0, "score_mae": 0.5297926624004965 }, "anli": { "count": 100, "accuracy": 0.63, "brier": 0.5421020492879414, "brier_soft": 0.5421020492879414, "nll": 0.9525621883553892, "nll_soft": 0.9525621883553892, "ece": 0.16033795682789762, "mean_confidence": 0.7903379568278975, "confident_error_rate": 0.08 }, "arc_challenge": { "count": 100, "accuracy": 0.94, "brier": 0.10067166282744358, "brier_soft": 0.10067166282744358, "nll": 0.1765881280453366, "nll_soft": 0.1765881280453366, "ece": 0.049984724800788485, "mean_confidence": 0.9541562154058013, "confident_error_rate": 0.01 }, "arc_easy": { "count": 100, "accuracy": 0.97, "brier": 0.04413162229222551, "brier_soft": 0.04413162229222551, "nll": 0.09823945269698767, "nll_soft": 0.09823945269698767, "ece": 0.032215481404806866, "mean_confidence": 0.9643697174383512, "confident_error_rate": 0.0 }, "banking77": { "count": 100, "accuracy": 0.75, "brier": 0.32482412936478505, "brier_soft": 0.32482412936478505, "nll": 0.7967586871695171, "nll_soft": 0.7967586871695171, "ece": 0.10943588766244959, "mean_confidence": 0.7685198614733331, "confident_error_rate": 0.01 }, "boolq": { "count": 125, "accuracy": 0.824, "brier": 0.2463948134922965, "brier_soft": 0.2463948134922965, "nll": 0.37807162862812616, "nll_soft": 0.37807162862812616, "ece": 0.062208085455247614, "mean_confidence": 0.8090593917787792, "confident_error_rate": 0.024, "evidence_auc": 1.0 }, "cb": { "count": 28, "accuracy": 1.0, "brier": 0.022647583217287822, "brier_soft": 0.022647583217287822, "nll": 0.08633507267229382, "nll_soft": 0.08633507267229382, "ece": 0.0780946990230204, "mean_confidence": 0.9219053009769796, "confident_error_rate": 0.0 }, "civil_comments_severity": { "count": 100, "accuracy": 0.86, "brier": 0.2456675530425123, "brier_soft": 0.1728906396682275, "nll": 0.503941587874954, "nll_soft": 0.638649591753842, "ece": 0.16160389502613104, "mean_confidence": 0.7384045083345092, "confident_error_rate": 0.0, "score_mae": 0.3738099519276761 }, "civil_comments_toxic": { "count": 100, "accuracy": 0.94, "brier": 0.11602869397484142, "brier_soft": 0.04566567351745008, "nll": 0.21750172983061675, "nll_soft": 0.29374434108311787, "ece": 0.10875554606138138, "mean_confidence": 0.8681357944166203, "confident_error_rate": 0.0 }, "clinc150": { "count": 100, "accuracy": 0.86, "brier": 0.22305660509274966, "brier_soft": 0.22305660509274966, "nll": 0.6025872624555403, "nll_soft": 0.6025872624555403, "ece": 0.09807589988409433, "mean_confidence": 0.8736587746039237, "confident_error_rate": 0.06 }, "cola": { "count": 100, "accuracy": 0.82, "brier": 0.2523992057795118, "brier_soft": 0.2523992057795118, "nll": 0.39240989420568745, "nll_soft": 0.39240989420568745, "ece": 0.05921553584016465, "mean_confidence": 0.8461464770037992, "confident_error_rate": 0.02 }, "commonsense_qa": { "count": 100, "accuracy": 0.85, "brier": 0.20557339122351911, "brier_soft": 0.20557339122351911, "nll": 0.37316502903223514, "nll_soft": 0.37316502903223514, "ece": 0.06105498864803392, "mean_confidence": 0.8862737420051069, "confident_error_rate": 0.03 }, "cosmos_qa": { "count": 120, "accuracy": 0.75, "brier": 0.36662150662163717, "brier_soft": 0.36662150662163717, "nll": 0.7134435721818425, "nll_soft": 0.7134435721818425, "ece": 0.1059972659018534, "mean_confidence": 0.8341453226399168, "confident_error_rate": 0.041666666666666664, "evidence_auc": 0.999375 }, "dates": { "count": 300, "accuracy": 0.9133333333333333, "brier": 0.13085868085462893, "brier_soft": 0.13085868085462893, "nll": 0.21953151134718468, "nll_soft": 0.21953151134718468, "ece": 0.041686146446700025, "mean_confidence": 0.8794316282366511, "confident_error_rate": 0.006666666666666667, "score_mae": 0.25286787804313066 }, "dbpedia": { "count": 100, "accuracy": 0.99, "brier": 0.0054752057851738724, "brier_soft": 0.0054752057851738724, "nll": 0.01716145009688713, "nll_soft": 0.01716145009688713, "ece": 0.014245544114448125, "mean_confidence": 0.9852099690835066, "confident_error_rate": 0.0 }, "dynasent": { "count": 100, "accuracy": 0.72, "brier": 0.41284497184536767, "brier_soft": 0.2900889097829367, "nll": 0.7092853395788147, "nll_soft": 0.7606595306372346, "ece": 0.19711298483299192, "mean_confidence": 0.5908801557778465, "confident_error_rate": 0.0, "score_mae": 0.4826536482310852 }, "enron_spam": { "count": 100, "accuracy": 0.98, "brier": 0.04113819214598982, "brier_soft": 0.04113819214598982, "nll": 0.08761406998679618, "nll_soft": 0.08761406998679618, "ece": 0.015903681566536102, "mean_confidence": 0.9741391053291784, "confident_error_rate": 0.02 }, "go_emotions": { "count": 100, "accuracy": 0.59, "brier": 0.6395508984210346, "brier_soft": 0.24193510902830995, "nll": 1.5803867250206407, "nll_soft": 2.0354749292162237, "ece": 0.22517856265972647, "mean_confidence": 0.36482143734027345, "confident_error_rate": 0.0 }, "hellaswag": { "count": 100, "accuracy": 0.93, "brier": 0.1078292724737536, "brier_soft": 0.1078292724737536, "nll": 0.20589254603888488, "nll_soft": 0.20589254603888488, "ece": 0.06967430987777687, "mean_confidence": 0.9220518174558482, "confident_error_rate": 0.0 }, "imdb": { "count": 100, "accuracy": 0.94, "brier": 0.07385005732170005, "brier_soft": 0.07385005732170005, "nll": 0.1359626492953412, "nll_soft": 0.1359626492953412, "ece": 0.04029954444016761, "mean_confidence": 0.9642720814018544, "confident_error_rate": 0.02 }, "ledgar": { "count": 100, "accuracy": 0.78, "brier": 0.3281153648327422, "brier_soft": 0.3281153648327422, "nll": 0.7741302141152906, "nll_soft": 0.7741302141152906, "ece": 0.0873475859579324, "mean_confidence": 0.7839166768781682, "confident_error_rate": 0.01 }, "liar": { "count": 100, "accuracy": 0.32, "brier": 0.7446807924003528, "brier_soft": 0.7446807924003528, "nll": 1.5117587745951337, "nll_soft": 1.5117587745951337, "ece": 0.08097392903128141, "mean_confidence": 0.2986578187328796, "confident_error_rate": 0.0, "score_mae": 1.036142943391838 }, "measuring_hate_speech": { "count": 100, "accuracy": 0.86, "brier": 0.3124419981758317, "brier_soft": 0.16390942252786078, "nll": 0.5875051718312956, "nll_soft": 0.6936316854447891, "ece": 0.2250050213791679, "mean_confidence": 0.6349949786208321, "confident_error_rate": 0.0, "score_mae": 0.6163252940584951 }, "medqa": { "count": 100, "accuracy": 0.65, "brier": 0.4597154624116973, "brier_soft": 0.4597154624116973, "nll": 0.8711963917666296, "nll_soft": 0.8711963917666296, "ece": 0.1116766692538281, "mean_confidence": 0.72033221601947, "confident_error_rate": 0.02 }, "mnli": { "count": 100, "accuracy": 0.93, "brier": 0.10529594164239192, "brier_soft": 0.10529594164239192, "nll": 0.2117929876710798, "nll_soft": 0.2117929876710798, "ece": 0.05925569835632346, "mean_confidence": 0.8817122608104968, "confident_error_rate": 0.01 }, "mrpc": { "count": 100, "accuracy": 0.81, "brier": 0.23545523687620534, "brier_soft": 0.23545523687620534, "nll": 0.36065976461182353, "nll_soft": 0.36065976461182353, "ece": 0.08009349827050247, "mean_confidence": 0.840161282202333, "confident_error_rate": 0.01 }, "multirc": { "count": 125, "accuracy": 0.712, "brier": 0.33425679170758704, "brier_soft": 0.33425679170758704, "nll": 0.47015119453748944, "nll_soft": 0.47015119453748944, "ece": 0.12508080340790598, "mean_confidence": 0.798369093709533, "confident_error_rate": 0.0, "evidence_auc": 0.9816 }, "openbookqa": { "count": 100, "accuracy": 0.89, "brier": 0.16850017640548665, "brier_soft": 0.16850017640548665, "nll": 0.3071681993046363, "nll_soft": 0.3071681993046363, "ece": 0.0681169423745702, "mean_confidence": 0.8801206851845036, "confident_error_rate": 0.0 }, "piqa": { "count": 100, "accuracy": 0.84, "brier": 0.22419783176420943, "brier_soft": 0.22419783176420943, "nll": 0.36462868341861826, "nll_soft": 0.36462868341861826, "ece": 0.08037792149433066, "mean_confidence": 0.87560922670375, "confident_error_rate": 0.05 }, "policy_application": { "count": 150, "accuracy": 0.9066666666666666, "brier": 0.1126946527275131, "brier_soft": 0.1126946527275131, "nll": 0.16724892277415848, "nll_soft": 0.16724892277415848, "ece": 0.0675511235449082, "mean_confidence": 0.9432567076192507, "confident_error_rate": 0.0, "evidence_auc": 1.0 }, "policy_hard": { "count": 300, "accuracy": 0.87, "brier": 0.1993095979231174, "brier_soft": 0.1993095979231174, "nll": 0.32845201396386725, "nll_soft": 0.32845201396386725, "ece": 0.05610869195965445, "mean_confidence": 0.9002004387015582, "confident_error_rate": 0.043333333333333335, "evidence_auc": 0.9759837962962963 }, "qqp": { "count": 100, "accuracy": 0.91, "brier": 0.14536355743868318, "brier_soft": 0.14536355743868318, "nll": 0.236511504425006, "nll_soft": 0.236511504425006, "ece": 0.06902310544783366, "mean_confidence": 0.863890066837007, "confident_error_rate": 0.0 }, "race": { "count": 125, "accuracy": 0.712, "brier": 0.33794214086609586, "brier_soft": 0.33794214086609586, "nll": 0.6353569038250544, "nll_soft": 0.6353569038250544, "ece": 0.11330167076929255, "mean_confidence": 0.8167852186501845, "confident_error_rate": 0.032, "evidence_auc": 0.9816 }, "rotten_tomatoes": { "count": 100, "accuracy": 0.95, "brier": 0.07359394082826959, "brier_soft": 0.07359394082826959, "nll": 0.12453968361648274, "nll_soft": 0.12453968361648274, "ece": 0.043066136326572566, "mean_confidence": 0.9365716100101855, "confident_error_rate": 0.01 }, "rte": { "count": 100, "accuracy": 0.91, "brier": 0.1378394419526845, "brier_soft": 0.1378394419526845, "nll": 0.22826701855192236, "nll_soft": 0.22826701855192236, "ece": 0.06151258102601688, "mean_confidence": 0.9404203748836127, "confident_error_rate": 0.03 }, "scotus": { "count": 100, "accuracy": 0.2, "brier": 0.8824845341223795, "brier_soft": 0.8824845341223795, "nll": 2.3725589448613023, "nll_soft": 2.3725589448613023, "ece": 0.03287297194524475, "mean_confidence": 0.2235796914649564, "confident_error_rate": 0.0 }, "shp": { "count": 100, "accuracy": 0.59, "brier": 0.4797397521419983, "brier_soft": 0.10692687814904282, "nll": 0.6725667723983078, "nll_soft": 0.6904620785786614, "ece": 0.0708668070573588, "mean_confidence": 0.539133192942641, "confident_error_rate": 0.0 }, "sms_spam": { "count": 100, "accuracy": 0.99, "brier": 0.01941783931224156, "brier_soft": 0.01941783931224156, "nll": 0.04414426991615675, "nll_soft": 0.04414426991615675, "ece": 0.023315330523776823, "mean_confidence": 0.9825703907007697, "confident_error_rate": 0.01 }, "snli": { "count": 100, "accuracy": 0.95, "brier": 0.08293153330688835, "brier_soft": 0.08293153330688835, "nll": 0.171101802859479, "nll_soft": 0.171101802859479, "ece": 0.04866259143581518, "mean_confidence": 0.9194179507114802, "confident_error_rate": 0.01 }, "social_iqa": { "count": 100, "accuracy": 0.84, "brier": 0.2572472931446341, "brier_soft": 0.2572472931446341, "nll": 0.4914033563092567, "nll_soft": 0.4914033563092567, "ece": 0.07708066338472419, "mean_confidence": 0.862248560054147, "confident_error_rate": 0.04 }, "sst2": { "count": 100, "accuracy": 0.97, "brier": 0.05812558128312039, "brier_soft": 0.05812558128312039, "nll": 0.12278980073652958, "nll_soft": 0.12278980073652958, "ece": 0.03465141864763422, "mean_confidence": 0.9609904656090619, "confident_error_rate": 0.02 }, "sst5": { "count": 100, "accuracy": 0.67, "brier": 0.4940931300179605, "brier_soft": 0.4940931300179605, "nll": 0.8709395956738571, "nll_soft": 0.8709395956738571, "ece": 0.1772210597159328, "mean_confidence": 0.49880477007066004, "confident_error_rate": 0.0, "score_mae": 0.43722177939425927 }, "stance_abortion": { "count": 60, "accuracy": 0.6333333333333333, "brier": 0.5098698490864988, "brier_soft": 0.5098698490864988, "nll": 0.8114548457536739, "nll_soft": 0.8114548457536739, "ece": 0.20824389335951496, "mean_confidence": 0.8298381203954579, "confident_error_rate": 0.06666666666666667 }, "stance_atheism": { "count": 60, "accuracy": 0.7666666666666667, "brier": 0.3626489360126296, "brier_soft": 0.3626489360126296, "nll": 0.6279343615381726, "nll_soft": 0.6279343615381726, "ece": 0.13401842415402584, "mean_confidence": 0.8509966632145957, "confident_error_rate": 0.05 }, "stance_feminist": { "count": 60, "accuracy": 0.7166666666666667, "brier": 0.36714295306448386, "brier_soft": 0.36714295306448386, "nll": 0.6301224389630461, "nll_soft": 0.6301224389630461, "ece": 0.09630169928292202, "mean_confidence": 0.7805527340063932, "confident_error_rate": 0.016666666666666666 }, "stance_hillary": { "count": 60, "accuracy": 0.6833333333333333, "brier": 0.43392563993709476, "brier_soft": 0.43392563993709476, "nll": 0.7006138108687522, "nll_soft": 0.7006138108687522, "ece": 0.14445342515385268, "mean_confidence": 0.7761897463072162, "confident_error_rate": 0.016666666666666666 }, "stsb": { "count": 100, "accuracy": 0.67, "brier": 0.47941600616535696, "brier_soft": 0.22487109798673896, "nll": 0.8814722777779251, "nll_soft": 0.938008268970549, "ece": 0.13853521132174568, "mean_confidence": 0.5474027968940396, "confident_error_rate": 0.0, "score_mae": 0.49143445522479867 }, "subjectivity": { "count": 100, "accuracy": 0.95, "brier": 0.060069737216477195, "brier_soft": 0.060069737216477195, "nll": 0.10706919888260735, "nll_soft": 0.10706919888260735, "ece": 0.047342527386868574, "mean_confidence": 0.9464740773328073, "confident_error_rate": 0.0 }, "swag": { "count": 100, "accuracy": 0.82, "brier": 0.26526580978725744, "brier_soft": 0.26526580978725744, "nll": 0.524827253241994, "nll_soft": 0.524827253241994, "ece": 0.07704899673765067, "mean_confidence": 0.8086559627591525, "confident_error_rate": 0.04 }, "trec": { "count": 100, "accuracy": 0.97, "brier": 0.06165980860279739, "brier_soft": 0.06165980860279739, "nll": 0.1440644009426513, "nll_soft": 0.1440644009426513, "ece": 0.034644018323219124, "mean_confidence": 0.953222112770253, "confident_error_rate": 0.01 }, "tweet_hate": { "count": 100, "accuracy": 0.59, "brier": 0.6099733449281726, "brier_soft": 0.6099733449281726, "nll": 0.9084071924004988, "nll_soft": 0.9084071924004988, "ece": 0.24263082120266097, "mean_confidence": 0.8118554592243875, "confident_error_rate": 0.14 }, "tweet_sentiment": { "count": 100, "accuracy": 0.64, "brier": 0.4367689299986772, "brier_soft": 0.4367689299986772, "nll": 0.7338407156604912, "nll_soft": 0.7338407156604912, "ece": 0.12957176644981752, "mean_confidence": 0.5765707988516637, "confident_error_rate": 0.0, "score_mae": 0.3922079327893301 }, "vitaminc": { "count": 100, "accuracy": 0.82, "brier": 0.2858661765372957, "brier_soft": 0.2858661765372957, "nll": 0.5383008018865654, "nll_soft": 0.5383008018865654, "ece": 0.097544971218811, "mean_confidence": 0.860535493801815, "confident_error_rate": 0.05 }, "wic": { "count": 100, "accuracy": 0.76, "brier": 0.3297731392352389, "brier_soft": 0.3297731392352389, "nll": 0.5004091591132902, "nll_soft": 0.5004091591132902, "ece": 0.05507502517150548, "mean_confidence": 0.7671851329825314, "confident_error_rate": 0.01 }, "winogrande": { "count": 100, "accuracy": 0.84, "brier": 0.24850296736454125, "brier_soft": 0.24850296736454125, "nll": 0.41686536630288906, "nll_soft": 0.41686536630288906, "ece": 0.07463382751134487, "mean_confidence": 0.8865171061775413, "confident_error_rate": 0.06 }, "yelp": { "count": 100, "accuracy": 0.69, "brier": 0.4359510518792043, "brier_soft": 0.4359510518792043, "nll": 0.7529076116320128, "nll_soft": 0.7529076116320128, "ece": 0.11329441101759959, "mean_confidence": 0.5767055889824003, "confident_error_rate": 0.0, "score_mae": 0.431973503589024 } }, "reliability": [ { "bin": "0.0-0.1", "count": 0, "confidence": NaN, "accuracy": NaN }, { "bin": "0.1-0.2", "count": 32, "confidence": 0.18775015679946586, "accuracy": 0.1875 }, { "bin": "0.2-0.3", "count": 167, "confidence": 0.23931850114374761, "accuracy": 0.2874251497005988 }, { "bin": "0.3-0.4", "count": 139, "confidence": 0.3597909667718615, "accuracy": 0.38848920863309355 }, { "bin": "0.4-0.5", "count": 441, "confidence": 0.4603512037898012, "accuracy": 0.5328798185941043 }, { "bin": "0.5-0.6", "count": 584, "confidence": 0.5552843800701477, "accuracy": 0.5941780821917808 }, { "bin": "0.6-0.7", "count": 443, "confidence": 0.6568538439207049, "accuracy": 0.6975169300225733 }, { "bin": "0.7-0.8", "count": 428, "confidence": 0.7512629337446123, "accuracy": 0.7289719626168224 }, { "bin": "0.8-0.9", "count": 658, "confidence": 0.8510231288751523, "accuracy": 0.8009118541033434 }, { "bin": "0.9-1.0", "count": 3221, "confidence": 0.972387448797977, "accuracy": 0.9630549518782987 } ], "seconds": 110.24949526786804, "model": "runs/assay-8b", "data": "data/v4/dev.jsonl" }