{ "meta": { "base_model": "Qwen/Qwen3.5-4B-Base", "d_head": 256, "lora": false, "renderer": { "max_state_tokens": 768, "max_instruction_tokens": 256, "max_option_tokens": 64 }, "step": 734, "val_nll": 0.27321311831474304 }, "temperatures": { "choice": 1.0929638744012644, "score": 1.1011288758276154, "noul": 1.228405177254357 }, "test": { "raw": { "overall": { "n": 5070, "acc": 0.8964496850967407, "nll": 0.276126503944397, "brier": 0.1475353091955185, "ece": 0.010809999694056847, "mean_conf": 0.9048414826393127, "score_mae": 0.41506025195121765 }, "by_type": { "choice": { "n": 3041, "acc": 0.9003617167472839, "nll": 0.2871852517127991, "brier": 0.1437402367591858, "ece": 0.013187985173783867, "mean_conf": 0.9091227054595947 }, "score": { "n": 273, "acc": 0.6849817037582397, "nll": 0.7681964039802551, "brier": 0.4400661289691925, "ece": 0.047672064850177796, "mean_conf": 0.6837503910064697, "score_mae": 0.41506025195121765 }, "noul": { "n": 1756, "acc": 0.9225512742996216, "nll": 0.18047469854354858, "brier": 0.10862862318754196, "ece": 0.013375402451816704, "mean_conf": 0.9318000674247742 } }, "by_source": { "ag_news": { "n": 86, "acc": 0.930232584476471, "nll": 0.23873434960842133, "brier": 0.10834771394729614, "ece": 0.056332710342662184, "mean_conf": 0.9181896448135376 }, "amazon_stars": { "n": 86, "acc": 0.5930232405662537, "nll": 0.8701136112213135, "brier": 0.5058151483535767, "ece": 0.10280728207642115, "mean_conf": 0.6330943703651428, "score_mae": 0.4689413011074066 }, "anli": { "n": 79, "acc": 0.746835470199585, "nll": 0.6076061129570007, "brier": 0.3763708174228668, "ece": 0.15590721699211704, "mean_conf": 0.8249932527542114 }, "banking77": { "n": 91, "acc": 0.9670329689979553, "nll": 0.109474316239357, "brier": 0.05170481652021408, "ece": 0.04274910372363963, "mean_conf": 0.9463766813278198 }, "boolq": { "n": 75, "acc": 0.9200000166893005, "nll": 0.22674815356731415, "brier": 0.14536462724208832, "ece": 0.07328386494371486, "mean_conf": 0.908048689365387 }, "cifar100": { "n": 297, "acc": 0.936026930809021, "nll": 0.19756226241588593, "brier": 0.096889927983284, "ece": 0.02602774308840307, "mean_conf": 0.9325385689735413 }, "civil_comments": { "n": 154, "acc": 0.9545454382896423, "nll": 0.1635725051164627, "brier": 0.08106027543544769, "ece": 0.08850163645923792, "mean_conf": 0.8801278471946716 }, "clinc_oos": { "n": 85, "acc": 0.9529411792755127, "nll": 0.10286867618560791, "brier": 0.057649165391922, "ece": 0.028260532641420116, "mean_conf": 0.9765596389770508 }, "cola": { "n": 89, "acc": 0.8202247023582458, "nll": 0.3538511097431183, "brier": 0.23637528717517853, "ece": 0.07856890498333513, "mean_conf": 0.8428974747657776 }, "commonsense_qa": { "n": 79, "acc": 0.797468364238739, "nll": 0.47439879179000854, "brier": 0.2571961283683777, "ece": 0.11401724499686991, "mean_conf": 0.8429151177406311 }, "copa": { "n": 19, "acc": 0.9473684430122375, "nll": 0.08234653621912003, "brier": 0.056282687932252884, "ece": 0.053942021972607, "mean_conf": 0.9678375720977783 }, "cub_birds": { "n": 209, "acc": 0.9282296895980835, "nll": 0.205092191696167, "brier": 0.09457369148731232, "ece": 0.0337111855618914, "mean_conf": 0.9249444007873535 }, "dbpedia": { "n": 83, "acc": 0.9879518151283264, "nll": 0.0267021507024765, "brier": 0.01387640181928873, "ece": 0.01773379953028531, "mean_conf": 0.9858810305595398 }, "dtd": { "n": 193, "acc": 0.9326424598693848, "nll": 0.1894713044166565, "brier": 0.10953276604413986, "ece": 0.0462280399812367, "mean_conf": 0.9360920786857605 }, "eurosat": { "n": 200, "acc": 0.9700000286102295, "nll": 0.09461035579442978, "brier": 0.04988160729408264, "ece": 0.022912378822553925, "mean_conf": 0.9720562696456909 }, "go_emotions": { "n": 156, "acc": 0.8461538553237915, "nll": 0.4075973331928253, "brier": 0.2239874303340912, "ece": 0.052582084089755954, "mean_conf": 0.8886739015579224 }, "gtsrb": { "n": 150, "acc": 0.9266666769981384, "nll": 0.2583926022052765, "brier": 0.10076691210269928, "ece": 0.0349075827405604, "mean_conf": 0.9396733641624451 }, "hellaswag": { "n": 71, "acc": 0.9718309640884399, "nll": 0.07258392125368118, "brier": 0.03423016890883446, "ece": 0.037721984350516546, "mean_conf": 0.9680467844009399 }, "imagenet100": { "n": 209, "acc": 0.9521530866622925, "nll": 0.1411520093679428, "brier": 0.06291362643241882, "ece": 0.01678104336341813, "mean_conf": 0.9670758247375488 }, "jailbreak": { "n": 43, "acc": 1.0, "nll": 0.003266148502007127, "brier": 0.0003119981265626848, "ece": 0.003183770989621637, "mean_conf": 0.9968161582946777 }, "mnli": { "n": 88, "acc": 0.9204545617103577, "nll": 0.17574532330036163, "brier": 0.10608046501874924, "ece": 0.039473747201574705, "mean_conf": 0.9106850028038025 }, "ms_marco": { "n": 77, "acc": 0.5714285969734192, "nll": 1.1672861576080322, "brier": 0.5488265156745911, "ece": 0.1069421570546632, "mean_conf": 0.5573300123214722 }, "newsgroups": { "n": 72, "acc": 0.8194444179534912, "nll": 0.656489372253418, "brier": 0.24557192623615265, "ece": 0.10580966421453894, "mean_conf": 0.8494588732719421 }, "openbookqa": { "n": 73, "acc": 0.9452054500579834, "nll": 0.2053627371788025, "brier": 0.10051807016134262, "ece": 0.07717989225561772, "mean_conf": 0.8888569474220276 }, "paws": { "n": 75, "acc": 0.8799999952316284, "nll": 0.252999871969223, "brier": 0.16616970300674438, "ece": 0.07703955284745612, "mean_conf": 0.957039475440979 }, "prompt_injection": { "n": 19, "acc": 1.0, "nll": 0.07026468217372894, "brier": 0.02881462872028351, "ece": 0.06134651247871403, "mean_conf": 0.9386534690856934 }, "qasc": { "n": 78, "acc": 0.8974359035491943, "nll": 0.22243788838386536, "brier": 0.11997074633836746, "ece": 0.058650574809634426, "mean_conf": 0.9030769467353821 }, "qnli": { "n": 79, "acc": 0.9620253443717957, "nll": 0.12711185216903687, "brier": 0.06058496609330177, "ece": 0.02656769830988215, "mean_conf": 0.9400109648704529 }, "qqp": { "n": 67, "acc": 0.8507462739944458, "nll": 0.32717540860176086, "brier": 0.20558716356754303, "ece": 0.05535818888270405, "mean_conf": 0.8827617168426514 }, "race": { "n": 71, "acc": 0.8169013857841492, "nll": 0.5218772292137146, "brier": 0.26671260595321655, "ece": 0.08260945939768694, "mean_conf": 0.868141770362854 }, "resisc45": { "n": 201, "acc": 0.9452736377716064, "nll": 0.17759159207344055, "brier": 0.07590000331401825, "ece": 0.03162269839005066, "mean_conf": 0.9709147810935974 }, "rte": { "n": 42, "acc": 0.9047619104385376, "nll": 0.21782515943050385, "brier": 0.12485634535551071, "ece": 0.05018191372233538, "mean_conf": 0.9342974424362183 }, "rvl_cdip": { "n": 207, "acc": 0.9227052927017212, "nll": 0.24853183329105377, "brier": 0.1186482235789299, "ece": 0.035717121188601514, "mean_conf": 0.9394121170043945 }, "sciq": { "n": 70, "acc": 0.9571428298950195, "nll": 0.08418832719326019, "brier": 0.05511568486690521, "ece": 0.05332319163331367, "mean_conf": 0.9662564992904663 }, "snli": { "n": 78, "acc": 0.9615384340286255, "nll": 0.13747791945934296, "brier": 0.07794130593538284, "ece": 0.06147045345163815, "mean_conf": 0.9255906343460083 }, "squad_v2": { "n": 83, "acc": 0.8795180916786194, "nll": 0.3885398805141449, "brier": 0.18468858301639557, "ece": 0.08604057614296412, "mean_conf": 0.9535071849822998 }, "sst5": { "n": 101, "acc": 0.6732673048973083, "nll": 0.8455938696861267, "brier": 0.4966614544391632, "ece": 0.10307709063944676, "mean_conf": 0.6207720041275024, "score_mae": 0.45541465282440186 }, "stanford_cars": { "n": 199, "acc": 0.9849246144294739, "nll": 0.06568868458271027, "brier": 0.033190734684467316, "ece": 0.02284979613965564, "mean_conf": 0.9779478907585144 }, "sun397": { "n": 205, "acc": 0.9658536314964294, "nll": 0.13169661164283752, "brier": 0.059237346053123474, "ece": 0.027937838926008762, "mean_conf": 0.9778258800506592 }, "toxic_chat": { "n": 103, "acc": 0.9805825352668762, "nll": 0.04486699774861336, "brier": 0.02695567160844803, "ece": 0.011975512422466195, "mean_conf": 0.9775879383087158 }, "tweet_irony": { "n": 73, "acc": 0.7397260069847107, "nll": 0.47673916816711426, "brier": 0.32497861981391907, "ece": 0.10002183849680613, "mean_conf": 0.8234904408454895 }, "tweet_offensive": { "n": 95, "acc": 0.800000011920929, "nll": 0.42309972643852234, "brier": 0.2671454846858978, "ece": 0.11630791231693371, "mean_conf": 0.8510119915008545 }, "tweet_sentiment": { "n": 96, "acc": 0.7604166865348816, "nll": 0.6009823679924011, "brier": 0.3709958493709564, "ece": 0.08339798031360968, "mean_conf": 0.7406583428382874 }, "vitaminc": { "n": 84, "acc": 0.9285714030265808, "nll": 0.20010635256767273, "brier": 0.11288775503635406, "ece": 0.055092808689364614, "mean_conf": 0.8885759711265564 }, "winogrande": { "n": 82, "acc": 0.8292682766914368, "nll": 0.3929581940174103, "brier": 0.2536102533340454, "ece": 0.09293118971498115, "mean_conf": 0.9037488698959351 }, "yahoo_topics": { "n": 71, "acc": 0.7746478915214539, "nll": 0.6088139414787292, "brier": 0.28299495577812195, "ece": 0.06163965007121763, "mean_conf": 0.7804079055786133 }, "yelp": { "n": 127, "acc": 0.834645688533783, "nll": 0.4275256395339966, "brier": 0.2361738681793213, "ece": 0.047331056411305306, "mean_conf": 0.858354926109314, "score_mae": 0.31378617882728577 } } }, "calibrated": { "overall": { "n": 5070, "acc": 0.8964496850967407, "nll": 0.2756565809249878, "brier": 0.14737455546855927, "ece": 0.0079650055739778, "mean_conf": 0.8922349810600281, "score_mae": 0.42143115401268005 }, "by_type": { "choice": { "n": 3041, "acc": 0.9003617167472839, "nll": 0.28574398159980774, "brier": 0.14335514605045319, "ece": 0.007999743385107894, "mean_conf": 0.8997042179107666 }, "score": { "n": 273, "acc": 0.6849817037582397, "nll": 0.7671821713447571, "brier": 0.439628005027771, "ece": 0.046648137280912105, "mean_conf": 0.6634480953216553, "score_mae": 0.42143115401268005 }, "noul": { "n": 1756, "acc": 0.9225512742996216, "nll": 0.18177150189876556, "brier": 0.10889948904514313, "ece": 0.019957571878301106, "mean_conf": 0.9148688316345215 } }, "by_source": { "ag_news": { "n": 86, "acc": 0.930232584476471, "nll": 0.2372349351644516, "brier": 0.10982950031757355, "ece": 0.06062877569098146, "mean_conf": 0.909740686416626 }, "amazon_stars": { "n": 86, "acc": 0.5930232405662537, "nll": 0.8707610964775085, "brier": 0.5034118294715881, "ece": 0.07147133660142413, "mean_conf": 0.6115179657936096, "score_mae": 0.4738217294216156 }, "anli": { "n": 79, "acc": 0.746835470199585, "nll": 0.5904105305671692, "brier": 0.3706202208995819, "ece": 0.12736005320452623, "mean_conf": 0.8066393733024597 }, "banking77": { "n": 91, "acc": 0.9670329689979553, "nll": 0.11517834663391113, "brier": 0.05284213647246361, "ece": 0.04865624303995164, "mean_conf": 0.9390164613723755 }, "boolq": { "n": 75, "acc": 0.9200000166893005, "nll": 0.230869323015213, "brier": 0.14464128017425537, "ece": 0.05709565622368751, "mean_conf": 0.8859748244285583 }, "cifar100": { "n": 297, "acc": 0.936026930809021, "nll": 0.19978056848049164, "brier": 0.09738589078187943, "ece": 0.03686069778541466, "mean_conf": 0.9233465194702148 }, "civil_comments": { "n": 154, "acc": 0.9545454382896423, "nll": 0.1945398449897766, "brier": 0.09554325044155121, "ece": 0.11094526093531587, "mean_conf": 0.8504656553268433 }, "clinc_oos": { "n": 85, "acc": 0.9529411792755127, "nll": 0.10213149338960648, "brier": 0.05607043579220772, "ece": 0.03434847114305911, "mean_conf": 0.9722728133201599 }, "cola": { "n": 89, "acc": 0.8202247023582458, "nll": 0.3592163622379303, "brier": 0.2345142364501953, "ece": 0.07689192332149841, "mean_conf": 0.8134418725967407 }, "commonsense_qa": { "n": 79, "acc": 0.797468364238739, "nll": 0.47140878438949585, "brier": 0.25501152873039246, "ece": 0.08131545514081495, "mean_conf": 0.8276560306549072 }, "copa": { "n": 19, "acc": 0.9473684430122375, "nll": 0.08365084230899811, "brier": 0.05513031780719757, "ece": 0.05683338684464534, "mean_conf": 0.9632882475852966 }, "cub_birds": { "n": 209, "acc": 0.9282296895980835, "nll": 0.20787712931632996, "brier": 0.09552589058876038, "ece": 0.03924435088578615, "mean_conf": 0.9157379269599915 }, "dbpedia": { "n": 83, "acc": 0.9879518151283264, "nll": 0.02914765290915966, "brier": 0.013987519778311253, "ece": 0.020033360317049276, "mean_conf": 0.9829169511795044 }, "dtd": { "n": 193, "acc": 0.9326424598693848, "nll": 0.19226685166358948, "brier": 0.10932108014822006, "ece": 0.04905586869892781, "mean_conf": 0.926359236240387 }, "eurosat": { "n": 200, "acc": 0.9700000286102295, "nll": 0.09493371099233627, "brier": 0.04979003965854645, "ece": 0.019272321772114326, "mean_conf": 0.964755117893219 }, "go_emotions": { "n": 156, "acc": 0.8461538553237915, "nll": 0.3998943567276001, "brier": 0.2204635888338089, "ece": 0.06369368834835576, "mean_conf": 0.8688168525695801 }, "gtsrb": { "n": 150, "acc": 0.9266666769981384, "nll": 0.25059908628463745, "brier": 0.09959184378385544, "ece": 0.04895997661024053, "mean_conf": 0.9357502460479736 }, "hellaswag": { "n": 71, "acc": 0.9718309640884399, "nll": 0.07610221207141876, "brier": 0.034769345074892044, "ece": 0.0420044008320708, "mean_conf": 0.9625624418258667 }, "imagenet100": { "n": 209, "acc": 0.9521530866622925, "nll": 0.140834778547287, "brier": 0.06210394203662872, "ece": 0.02685701278059019, "mean_conf": 0.9609060883522034 }, "jailbreak": { "n": 43, "acc": 1.0, "nll": 0.007107510231435299, "brier": 0.0007506122929044068, "ece": 0.006904559023125811, "mean_conf": 0.9930953979492188 }, "mnli": { "n": 88, "acc": 0.9204545617103577, "nll": 0.18531149625778198, "brier": 0.10737426578998566, "ece": 0.06684734037705664, "mean_conf": 0.8966826796531677 }, "ms_marco": { "n": 77, "acc": 0.5714285969734192, "nll": 1.1738694906234741, "brier": 0.5511745810508728, "ece": 0.1849330931448978, "mean_conf": 0.5303053855895996 }, "newsgroups": { "n": 72, "acc": 0.8194444179534912, "nll": 0.6413953304290771, "brier": 0.2439313530921936, "ece": 0.0672152123840676, "mean_conf": 0.8347481489181519 }, "openbookqa": { "n": 73, "acc": 0.9452054500579834, "nll": 0.21645799279212952, "brier": 0.10464119911193848, "ece": 0.07754900295525902, "mean_conf": 0.8753600716590881 }, "paws": { "n": 75, "acc": 0.8799999952316284, "nll": 0.23008832335472107, "brier": 0.15555186569690704, "ece": 0.07505809075017267, "mean_conf": 0.9440792798995972 }, "prompt_injection": { "n": 19, "acc": 1.0, "nll": 0.0893387570977211, "brier": 0.0387185737490654, "ece": 0.07711471421890259, "mean_conf": 0.9228852987289429 }, "qasc": { "n": 78, "acc": 0.8974359035491943, "nll": 0.2294393628835678, "brier": 0.12008954584598541, "ece": 0.07760370314998621, "mean_conf": 0.8928042650222778 }, "qnli": { "n": 79, "acc": 0.9620253443717957, "nll": 0.13672247529029846, "brier": 0.0658198893070221, "ece": 0.04174679779133661, "mean_conf": 0.9221668243408203 }, "qqp": { "n": 67, "acc": 0.8507462739944458, "nll": 0.3204824924468994, "brier": 0.20365533232688904, "ece": 0.05547176195417913, "mean_conf": 0.8596541285514832 }, "race": { "n": 71, "acc": 0.8169013857841492, "nll": 0.508931040763855, "brier": 0.2651640772819519, "ece": 0.09953184393917758, "mean_conf": 0.855144202709198 }, "resisc45": { "n": 201, "acc": 0.9452736377716064, "nll": 0.1719132363796234, "brier": 0.07461114972829819, "ece": 0.03276641192363214, "mean_conf": 0.9654509425163269 }, "rte": { "n": 42, "acc": 0.9047619104385376, "nll": 0.2149493396282196, "brier": 0.12409208714962006, "ece": 0.09782490985591286, "mean_conf": 0.9137647747993469 }, "rvl_cdip": { "n": 207, "acc": 0.9227052927017212, "nll": 0.24763381481170654, "brier": 0.11820948868989944, "ece": 0.024818566785942367, "mean_conf": 0.9276553392410278 }, "sciq": { "n": 70, "acc": 0.9571428298950195, "nll": 0.08641297370195389, "brier": 0.0542028583586216, "ece": 0.05665828745125506, "mean_conf": 0.9611032605171204 }, "snli": { "n": 78, "acc": 0.9615384340286255, "nll": 0.14562663435935974, "brier": 0.07991129904985428, "ece": 0.05743888654096996, "mean_conf": 0.9150496125221252 }, "squad_v2": { "n": 83, "acc": 0.8795180916786194, "nll": 0.3431330621242523, "brier": 0.18107561767101288, "ece": 0.08313956865378169, "mean_conf": 0.9385043978691101 }, "sst5": { "n": 101, "acc": 0.6732673048973083, "nll": 0.8503657579421997, "brier": 0.49845677614212036, "ece": 0.0914942533168388, "mean_conf": 0.5979520082473755, "score_mae": 0.46103909611701965 }, "stanford_cars": { "n": 199, "acc": 0.9849246144294739, "nll": 0.0664161965250969, "brier": 0.03385894373059273, "ece": 0.02054634083136521, "mean_conf": 0.9744680523872375 }, "sun397": { "n": 205, "acc": 0.9658536314964294, "nll": 0.12713490426540375, "brier": 0.05907195806503296, "ece": 0.03913777098139376, "mean_conf": 0.973029613494873 }, "toxic_chat": { "n": 103, "acc": 0.9805825352668762, "nll": 0.04893926531076431, "brier": 0.027793018147349358, "ece": 0.0175806231729555, "mean_conf": 0.971018373966217 }, "tweet_irony": { "n": 73, "acc": 0.7397260069847107, "nll": 0.46547093987464905, "brier": 0.3147864043712616, "ece": 0.1386445264786409, "mean_conf": 0.7930445075035095 }, "tweet_offensive": { "n": 95, "acc": 0.800000011920929, "nll": 0.4134097099304199, "brier": 0.263822078704834, "ece": 0.12641781317580403, "mean_conf": 0.8215466737747192 }, "tweet_sentiment": { "n": 96, "acc": 0.7604166865348816, "nll": 0.6009461283683777, "brier": 0.37088480591773987, "ece": 0.09440179414827354, "mean_conf": 0.7233335375785828 }, "vitaminc": { "n": 84, "acc": 0.9285714030265808, "nll": 0.21254083514213562, "brier": 0.11658700555562973, "ece": 0.060456716703195634, "mean_conf": 0.87336665391922 }, "winogrande": { "n": 82, "acc": 0.8292682766914368, "nll": 0.3811376690864563, "brier": 0.24891741573810577, "ece": 0.07569023195409919, "mean_conf": 0.8934732675552368 }, "yahoo_topics": { "n": 71, "acc": 0.7746478915214539, "nll": 0.6164750456809998, "brier": 0.2843913733959198, "ece": 0.10321198388129582, "mean_conf": 0.7571716904640198 }, "yelp": { "n": 127, "acc": 0.834645688533783, "nll": 0.4197368919849396, "brier": 0.23375587165355682, "ece": 0.036530637687290035, "mean_conf": 0.8449493646621704, "score_mae": 0.32252421975135803 } } }, "reliability_raw": [ { "lo": 0.133, "hi": 0.2, "conf": 0.17327187955379486, "acc": 0.0, "frac": 0.0007889546104706824 }, { "lo": 0.2, "hi": 0.267, "conf": 0.23755492269992828, "acc": 0.2222222238779068, "frac": 0.001775147975422442 }, { "lo": 0.267, "hi": 0.333, "conf": 0.3051908314228058, "acc": 0.2800000011920929, "frac": 0.004930966533720493 }, { "lo": 0.333, "hi": 0.4, "conf": 0.36935192346572876, "acc": 0.26923078298568726, "frac": 0.0051282052882015705 }, { "lo": 0.4, "hi": 0.467, "conf": 0.43677520751953125, "acc": 0.42307692766189575, "frac": 0.010256410576403141 }, { "lo": 0.467, "hi": 0.533, "conf": 0.5065982937812805, "acc": 0.5289855003356934, "frac": 0.027218934148550034 }, { "lo": 0.533, "hi": 0.6, "conf": 0.5669992566108704, "acc": 0.5779221057891846, "frac": 0.03037475422024727 }, { "lo": 0.6, "hi": 0.667, "conf": 0.63744056224823, "acc": 0.6171428561210632, "frac": 0.03451676666736603 }, { "lo": 0.667, "hi": 0.733, "conf": 0.7005413770675659, "acc": 0.6280487775802612, "frac": 0.03234713897109032 }, { "lo": 0.733, "hi": 0.8, "conf": 0.767250657081604, "acc": 0.7511312365531921, "frac": 0.043589744716882706 }, { "lo": 0.8, "hi": 0.867, "conf": 0.834570586681366, "acc": 0.80694979429245, "frac": 0.051084812730550766 }, { "lo": 0.867, "hi": 0.933, "conf": 0.9034878015518188, "acc": 0.9073033928871155, "frac": 0.07021696120500565 }, { "lo": 0.933, "hi": 1.0, "conf": 0.9904627203941345, "acc": 0.9853742718696594, "frac": 0.687771201133728 } ], "reliability_calibrated": [ { "lo": 0.133, "hi": 0.2, "conf": 0.17058970034122467, "acc": 0.0, "frac": 0.0009861933067440987 }, { "lo": 0.2, "hi": 0.267, "conf": 0.24141208827495575, "acc": 0.23076923191547394, "frac": 0.0025641026441007853 }, { "lo": 0.267, "hi": 0.333, "conf": 0.30261722207069397, "acc": 0.3076923191547394, "frac": 0.0051282052882015705 }, { "lo": 0.333, "hi": 0.4, "conf": 0.3718169331550598, "acc": 0.29032257199287415, "frac": 0.006114398594945669 }, { "lo": 0.4, "hi": 0.467, "conf": 0.4343971312046051, "acc": 0.5079365372657776, "frac": 0.012426035478711128 }, { "lo": 0.467, "hi": 0.533, "conf": 0.5043900012969971, "acc": 0.5302013158798218, "frac": 0.029388559982180595 }, { "lo": 0.533, "hi": 0.6, "conf": 0.5663641691207886, "acc": 0.5602409839630127, "frac": 0.032741617411375046 }, { "lo": 0.6, "hi": 0.667, "conf": 0.6334572434425354, "acc": 0.6188119053840637, "frac": 0.039842210710048676 }, { "lo": 0.667, "hi": 0.733, "conf": 0.7039550542831421, "acc": 0.6941747665405273, "frac": 0.040631163865327835 }, { "lo": 0.733, "hi": 0.8, "conf": 0.7678379416465759, "acc": 0.7991452813148499, "frac": 0.04615384712815285 }, { "lo": 0.8, "hi": 0.867, "conf": 0.8343891501426697, "acc": 0.8534798622131348, "frac": 0.05384615436196327 }, { "lo": 0.867, "hi": 0.933, "conf": 0.9034571051597595, "acc": 0.9114832282066345, "frac": 0.08244576305150986 }, { "lo": 0.933, "hi": 1.0, "conf": 0.9883154630661011, "acc": 0.9902557730674744, "frac": 0.6477317810058594 } ] }, "heldout": { "raw": { "overall": { "n": 9888, "acc": 0.7774069309234619, "nll": 0.6259722709655762, "brier": 0.3122439980506897, "ece": 0.055361537657564505, "mean_conf": 0.8327684998512268 }, "by_type": { "choice": { "n": 9088, "acc": 0.7644146084785461, "nll": 0.6617501974105835, "brier": 0.3292696177959442, "ece": 0.05688635497695306, "mean_conf": 0.8213009238243103 }, "noul": { "n": 800, "acc": 0.925000011920929, "nll": 0.21953491866588593, "brier": 0.11883323639631271, "ece": 0.042017475422659586, "mean_conf": 0.9630394577980042 } }, "by_source": { "arc_easy": { "n": 500, "acc": 0.9760000109672546, "nll": 0.10126849263906479, "brier": 0.04353426396846771, "ece": 0.03141405401313018, "mean_conf": 0.9497805237770081 }, "askubuntu": { "n": 359, "acc": 0.5181058645248413, "nll": 1.715165376663208, "brier": 0.6542366743087769, "ece": 0.16121354500097346, "mean_conf": 0.679319441318512 }, "cifar10": { "n": 300, "acc": 0.9766666889190674, "nll": 0.06821396946907043, "brier": 0.033433299511671066, "ece": 0.0244025032659336, "mean_conf": 0.9716125726699829 }, "emotion": { "n": 500, "acc": 0.5799999833106995, "nll": 1.305448293685913, "brier": 0.6054953932762146, "ece": 0.1998288362110465, "mean_conf": 0.7798287868499756 }, "food101": { "n": 300, "acc": 0.8766666650772095, "nll": 0.35791701078414917, "brier": 0.15910951793193817, "ece": 0.04321668811345984, "mean_conf": 0.8755965232849121 }, "imdb": { "n": 500, "acc": 0.9399999976158142, "nll": 0.15117090940475464, "brier": 0.0857759416103363, "ece": 0.033096101656337, "mean_conf": 0.9650362730026245 }, "massive_intent": { "n": 500, "acc": 0.7459999918937683, "nll": 0.9858649969100952, "brier": 0.3682084381580353, "ece": 0.09032174787909839, "mean_conf": 0.8352954983711243 }, "mnist": { "n": 300, "acc": 0.95333331823349, "nll": 0.15937867760658264, "brier": 0.06808914244174957, "ece": 0.03201363153018352, "mean_conf": 0.9333906769752502 }, "ni_unseen_tasks": { "n": 5529, "acc": 0.752215564250946, "nll": 0.6323504447937012, "brier": 0.34408876299858093, "ece": 0.05498220219487805, "mean_conf": 0.8071978688240051 }, "pets": { "n": 300, "acc": 0.8500000238418579, "nll": 0.5367151498794556, "brier": 0.22500383853912354, "ece": 0.06403641861143206, "mean_conf": 0.8197709918022156 }, "pope": { "n": 300, "acc": 0.8999999761581421, "nll": 0.33347487449645996, "brier": 0.17392873764038086, "ece": 0.07330050353340761, "mean_conf": 0.9597114324569702 }, "pubmed_qa": { "n": 500, "acc": 0.7080000042915344, "nll": 0.738036036491394, "brier": 0.4015096426010132, "ece": 0.08785723579766086, "mean_conf": 0.7890949249267578 } } }, "calibrated": { "overall": { "n": 9888, "acc": 0.7774069309234619, "nll": 0.6107043027877808, "brier": 0.30924949049949646, "ece": 0.040644236667579256, "mean_conf": 0.8180511593818665 }, "by_type": { "choice": { "n": 9088, "acc": 0.7644146084785461, "nll": 0.6470264792442322, "brier": 0.32634666562080383, "ece": 0.041729311528364656, "mean_conf": 0.8061439394950867 }, "noul": { "n": 800, "acc": 0.925000011920929, "nll": 0.1980842649936676, "brier": 0.11502549052238464, "ece": 0.03511890266276535, "mean_conf": 0.9533174633979797 } }, "by_source": { "arc_easy": { "n": 500, "acc": 0.9760000109672546, "nll": 0.10715165734291077, "brier": 0.04551118239760399, "ece": 0.03949616313766471, "mean_conf": 0.9417195320129395 }, "askubuntu": { "n": 359, "acc": 0.5181058645248413, "nll": 1.6524147987365723, "brier": 0.6454375982284546, "ece": 0.1338557621591943, "mean_conf": 0.651961624622345 }, "cifar10": { "n": 300, "acc": 0.9766666889190674, "nll": 0.07086114585399628, "brier": 0.03372351825237274, "ece": 0.03409448859720851, "mean_conf": 0.9671555757522583 }, "emotion": { "n": 500, "acc": 0.5799999833106995, "nll": 1.2474008798599243, "brier": 0.5944139957427979, "ece": 0.17934864653363677, "mean_conf": 0.7593486309051514 }, "food101": { "n": 300, "acc": 0.8766666650772095, "nll": 0.3655674159526825, "brier": 0.1599694937467575, "ece": 0.05726011354152369, "mean_conf": 0.8594070672988892 }, "imdb": { "n": 500, "acc": 0.9399999976158142, "nll": 0.1412155032157898, "brier": 0.08349451422691345, "ece": 0.025615910650256563, "mean_conf": 0.955990195274353 }, "massive_intent": { "n": 500, "acc": 0.7459999918937683, "nll": 0.9505149722099304, "brier": 0.36250603199005127, "ece": 0.07815012133194074, "mean_conf": 0.8151710033416748 }, "mnist": { "n": 300, "acc": 0.95333331823349, "nll": 0.1676100343465805, "brier": 0.07024406641721725, "ece": 0.044357041122790236, "mean_conf": 0.9220753908157349 }, "ni_unseen_tasks": { "n": 5529, "acc": 0.752215564250946, "nll": 0.6206715106964111, "brier": 0.34118887782096863, "ece": 0.040373205554264546, "mean_conf": 0.7925888299942017 }, "pets": { "n": 300, "acc": 0.8500000238418579, "nll": 0.5368061065673828, "brier": 0.22789603471755981, "ece": 0.07454185713334927, "mean_conf": 0.8035118579864502 }, "pope": { "n": 300, "acc": 0.8999999761581421, "nll": 0.29286548495292664, "brier": 0.1675771325826645, "ece": 0.06612932776747837, "mean_conf": 0.9488630890846252 }, "pubmed_qa": { "n": 500, "acc": 0.7080000042915344, "nll": 0.7209590077400208, "brier": 0.39785510301589966, "ece": 0.07103049902030684, "mean_conf": 0.7723875641822815 } } }, "reliability_raw": [ { "lo": 0.133, "hi": 0.2, "conf": 0.18254680931568146, "acc": 0.0, "frac": 0.00040453075780533254 }, { "lo": 0.2, "hi": 0.267, "conf": 0.24415171146392822, "acc": 0.125, "frac": 0.004045307636260986 }, { "lo": 0.267, "hi": 0.333, "conf": 0.3000350594520569, "acc": 0.2711864411830902, "frac": 0.011933657340705395 }, { "lo": 0.333, "hi": 0.4, "conf": 0.36993393301963806, "acc": 0.2829268276691437, "frac": 0.020732199773192406 }, { "lo": 0.4, "hi": 0.467, "conf": 0.434598833322525, "acc": 0.356890469789505, "frac": 0.02862055040895939 }, { "lo": 0.467, "hi": 0.533, "conf": 0.5052576661109924, "acc": 0.4368070960044861, "frac": 0.04561084136366844 }, { "lo": 0.533, "hi": 0.6, "conf": 0.5669507384300232, "acc": 0.4856114983558655, "frac": 0.05622977390885353 }, { "lo": 0.6, "hi": 0.667, "conf": 0.6334361433982849, "acc": 0.5653104782104492, "frac": 0.047228965908288956 }, { "lo": 0.667, "hi": 0.733, "conf": 0.701598048210144, "acc": 0.6368821263313293, "frac": 0.05319579318165779 }, { "lo": 0.733, "hi": 0.8, "conf": 0.7661896347999573, "acc": 0.7110726833343506, "frac": 0.058454692363739014 }, { "lo": 0.8, "hi": 0.867, "conf": 0.8357764482498169, "acc": 0.7466007471084595, "frac": 0.08181634545326233 }, { "lo": 0.867, "hi": 0.933, "conf": 0.9009120464324951, "acc": 0.8152466416358948, "frac": 0.1127629429101944 }, { "lo": 0.933, "hi": 1.0, "conf": 0.9835764169692993, "acc": 0.9503800868988037, "frac": 0.4789643883705139 } ], "reliability_calibrated": [ { "lo": 0.133, "hi": 0.2, "conf": 0.17538046836853027, "acc": 0.0, "frac": 0.0006067961221560836 }, { "lo": 0.2, "hi": 0.267, "conf": 0.24302583932876587, "acc": 0.16981132328510284, "frac": 0.005360032431781292 }, { "lo": 0.267, "hi": 0.333, "conf": 0.30060428380966187, "acc": 0.25362318754196167, "frac": 0.013956310227513313 }, { "lo": 0.333, "hi": 0.4, "conf": 0.36858007311820984, "acc": 0.3234042525291443, "frac": 0.023766180500388145 }, { "lo": 0.4, "hi": 0.467, "conf": 0.4340149462223053, "acc": 0.3523809611797333, "frac": 0.03185679763555527 }, { "lo": 0.467, "hi": 0.533, "conf": 0.5055366158485413, "acc": 0.45656564831733704, "frac": 0.05006067827343941 }, { "lo": 0.533, "hi": 0.6, "conf": 0.5651605129241943, "acc": 0.5167548656463623, "frac": 0.05734223127365112 }, { "lo": 0.6, "hi": 0.667, "conf": 0.6337770223617554, "acc": 0.5630081295967102, "frac": 0.04975728318095207 }, { "lo": 0.667, "hi": 0.733, "conf": 0.7002703547477722, "acc": 0.6780238747596741, "frac": 0.059364885091781616 }, { "lo": 0.733, "hi": 0.8, "conf": 0.7682633399963379, "acc": 0.7181528806686401, "frac": 0.06351132690906525 }, { "lo": 0.8, "hi": 0.867, "conf": 0.8352631330490112, "acc": 0.7694915533065796, "frac": 0.08950242400169373 }, { "lo": 0.867, "hi": 0.933, "conf": 0.9024354219436646, "acc": 0.8416013717651367, "frac": 0.11620145291090012 }, { "lo": 0.933, "hi": 1.0, "conf": 0.9815592169761658, "acc": 0.9596588015556335, "frac": 0.43871358036994934 } ] }, "fork_invariance_max_dp": 6.145238876342773e-05, "latency_ms": { "1": { "fork": 108.40368270874023, "separate": 959.8970413208008 }, "4": { "fork": 107.42926597595215, "separate": 1998.2314109802246 }, "16": { "fork": 113.10410499572754, "separate": 7827.5275230407715 }, "64": { "fork": 256.7019462585449, "separate": 16031.468868255615 } } }