{ "overall": { "count": 6113, "accuracy": 0.7906101750368068, "brier": 0.2905109878230025, "brier_soft": 0.26728262886573473, "nll": 0.5449381413787723, "nll_soft": 0.5611535112867304, "ece": 0.045661348853598224, "mean_confidence": 0.8293620655723313, "confident_error_rate": 0.032717160150498935, "score_mae": 0.5152794260773468, "evidence_auc": 0.9997787857086019 }, "by_type": { "bool": { "count": 1898, "accuracy": 0.8535300316122234, "brier": 0.20983104954609286, "brier_soft": 0.20647029718217128, "nll": 0.34005517640404503, "nll_soft": 0.3452724980816234, "ece": 0.058970879163590936, "mean_confidence": 0.9125009107758142, "confident_error_rate": 0.04056902002107481, "evidence_auc": 0.9991618000687049 }, "choice": { "count": 3248, "accuracy": 0.7924876847290641, "brier": 0.29241979456448064, "brier_soft": 0.26966333263184994, "nll": 0.5891846494040984, "nll_soft": 0.6056750666591111, "ece": 0.056842395140909024, "mean_confidence": 0.8473377443208175, "confident_error_rate": 0.037561576354679806, "evidence_auc": 0.999926113573992 }, "score": { "count": 967, "accuracy": 0.6608066184074457, "brier": 0.4424558880848984, "brier_soft": 0.3786469305235023, "nll": 0.7984595575688168, "nll_soft": 0.8353377421179621, "ece": 0.07923403877747387, "mean_confidence": 0.6058020523652026, "confident_error_rate": 0.001034126163391934, "score_mae": 0.5152794260773468 } }, "by_source": { "ag_news": { "count": 100, "accuracy": 0.91, "brier": 0.1457681115044795, "brier_soft": 0.1457681115044795, "nll": 0.2806835892729357, "nll_soft": 0.2806835892729357, "ece": 0.06551793024708885, "mean_confidence": 0.9522883211811144, "confident_error_rate": 0.05 }, "amazon_polarity": { "count": 100, "accuracy": 0.93, "brier": 0.09681691910594509, "brier_soft": 0.09681691910594509, "nll": 0.14046548898129674, "nll_soft": 0.14046548898129674, "ece": 0.052496080969926015, "mean_confidence": 0.9707786674013664, "confident_error_rate": 0.01 }, "amazon_stars": { "count": 100, "accuracy": 0.66, "brier": 0.5044599603848816, "brier_soft": 0.5044599603848816, "nll": 0.869633277428378, "nll_soft": 0.869633277428378, "ece": 0.11863087715317772, "mean_confidence": 0.576783671003821, "confident_error_rate": 0.0, "score_mae": 0.49169127512149585 }, "anli": { "count": 100, "accuracy": 0.58, "brier": 0.6073661288342754, "brier_soft": 0.6073661288342754, "nll": 1.0969887313858857, "nll_soft": 1.0969887313858857, "ece": 0.2611784174019323, "mean_confidence": 0.8288514520699011, "confident_error_rate": 0.11 }, "arc_challenge": { "count": 100, "accuracy": 0.92, "brier": 0.14429682116011844, "brier_soft": 0.14429682116011844, "nll": 0.2640587153943526, "nll_soft": 0.2640587153943526, "ece": 0.0850214039559398, "mean_confidence": 0.9523708441098319, "confident_error_rate": 0.04 }, "arc_easy": { "count": 100, "accuracy": 0.96, "brier": 0.06183016851108757, "brier_soft": 0.06183016851108757, "nll": 0.12801979988318146, "nll_soft": 0.12801979988318146, "ece": 0.05157003044933564, "mean_confidence": 0.9644089935643387, "confident_error_rate": 0.0 }, "banking77": { "count": 100, "accuracy": 0.73, "brier": 0.37647411046583196, "brier_soft": 0.37647411046583196, "nll": 0.967146808339886, "nll_soft": 0.967146808339886, "ece": 0.10248049282910536, "mean_confidence": 0.8103718617199586, "confident_error_rate": 0.04 }, "boolq": { "count": 125, "accuracy": 0.768, "brier": 0.3015155723338054, "brier_soft": 0.3015155723338054, "nll": 0.4403017247566234, "nll_soft": 0.4403017247566234, "ece": 0.0691713155356551, "mean_confidence": 0.8214496526402538, "confident_error_rate": 0.008, "evidence_auc": 1.0 }, "cb": { "count": 28, "accuracy": 0.9285714285714286, "brier": 0.0885260800587447, "brier_soft": 0.0885260800587447, "nll": 0.1389439819866921, "nll_soft": 0.1389439819866921, "ece": 0.07268257534162591, "mean_confidence": 0.9543104112884626, "confident_error_rate": 0.0 }, "civil_comments_severity": { "count": 100, "accuracy": 0.84, "brier": 0.26092584575590455, "brier_soft": 0.18027036778246955, "nll": 0.5385317230811851, "nll_soft": 0.6599217016387524, "ece": 0.10881889299660845, "mean_confidence": 0.7390225605257076, "confident_error_rate": 0.01, "score_mae": 0.3756541661037601 }, "civil_comments_toxic": { "count": 100, "accuracy": 0.93, "brier": 0.1116733017198554, "brier_soft": 0.04788622185262424, "nll": 0.20032056546099006, "nll_soft": 0.299345330901428, "ece": 0.05349609414767832, "mean_confidence": 0.9194762200522314, "confident_error_rate": 0.02 }, "clinc150": { "count": 100, "accuracy": 0.87, "brier": 0.22371522830971066, "brier_soft": 0.22371522830971066, "nll": 0.7134262722233493, "nll_soft": 0.7134262722233493, "ece": 0.09005755998184956, "mean_confidence": 0.9235783766174436, "confident_error_rate": 0.05 }, "cola": { "count": 100, "accuracy": 0.75, "brier": 0.3315955099695127, "brier_soft": 0.3315955099695127, "nll": 0.5199970361758424, "nll_soft": 0.5199970361758424, "ece": 0.14225371716410648, "mean_confidence": 0.8922537171641064, "confident_error_rate": 0.08 }, "commonsense_qa": { "count": 100, "accuracy": 0.87, "brier": 0.2043245094910104, "brier_soft": 0.2043245094910104, "nll": 0.3885151902545537, "nll_soft": 0.3885151902545537, "ece": 0.07296244357028275, "mean_confidence": 0.9304146401653779, "confident_error_rate": 0.04 }, "cosmos_qa": { "count": 120, "accuracy": 0.7333333333333333, "brier": 0.4125665910418836, "brier_soft": 0.4125665910418836, "nll": 0.8681569441961412, "nll_soft": 0.8681569441961412, "ece": 0.13699584744487603, "mean_confidence": 0.8332889952145706, "confident_error_rate": 0.06666666666666667, "evidence_auc": 0.9990625 }, "dates": { "count": 300, "accuracy": 0.9066666666666666, "brier": 0.1324410444899149, "brier_soft": 0.1324410444899149, "nll": 0.21817804469288452, "nll_soft": 0.21817804469288452, "ece": 0.03892548374314649, "mean_confidence": 0.8972209231665249, "confident_error_rate": 0.013333333333333334, "score_mae": 0.2506003548185511 }, "dbpedia": { "count": 100, "accuracy": 1.0, "brier": 1.6184818300542076e-05, "brier_soft": 1.6184818300542076e-05, "nll": 0.001827317797337679, "nll_soft": 0.001827317797337679, "ece": 0.0018219140015888202, "mean_confidence": 0.9981780859984112, "confident_error_rate": 0.0 }, "dynasent": { "count": 100, "accuracy": 0.69, "brier": 0.4149544413021475, "brier_soft": 0.2949069170758868, "nll": 0.70767730702296, "nll_soft": 0.7696773071306522, "ece": 0.1306706361394352, "mean_confidence": 0.6253540882792548, "confident_error_rate": 0.0, "score_mae": 0.47241350842045404 }, "enron_spam": { "count": 100, "accuracy": 0.95, "brier": 0.06551335557649704, "brier_soft": 0.06551335557649704, "nll": 0.10981263791638528, "nll_soft": 0.10981263791638528, "ece": 0.03485533983225947, "mean_confidence": 0.9848553398322596, "confident_error_rate": 0.03 }, "go_emotions": { "count": 100, "accuracy": 0.54, "brier": 0.6194861171793581, "brier_soft": 0.2513976651604882, "nll": 1.5335175515779613, "nll_soft": 2.049709515080812, "ece": 0.13643982953461417, "mean_confidence": 0.4588293132172613, "confident_error_rate": 0.0 }, "hellaswag": { "count": 100, "accuracy": 0.88, "brier": 0.1787256334271709, "brier_soft": 0.1787256334271709, "nll": 0.343730914235669, "nll_soft": 0.343730914235669, "ece": 0.07519650793085166, "mean_confidence": 0.9438893610557416, "confident_error_rate": 0.04 }, "imdb": { "count": 100, "accuracy": 0.94, "brier": 0.09857211855143005, "brier_soft": 0.09857211855143005, "nll": 0.16853336129576013, "nll_soft": 0.16853336129576013, "ece": 0.05390522915327187, "mean_confidence": 0.9718127557710655, "confident_error_rate": 0.03 }, "ledgar": { "count": 100, "accuracy": 0.75, "brier": 0.36107508218738843, "brier_soft": 0.36107508218738843, "nll": 0.9228652156026649, "nll_soft": 0.9228652156026649, "ece": 0.12147796710352471, "mean_confidence": 0.8216070461871731, "confident_error_rate": 0.04 }, "liar": { "count": 100, "accuracy": 0.24, "brier": 0.7720101279378543, "brier_soft": 0.7720101279378543, "nll": 1.5780722835852494, "nll_soft": 1.5780722835852494, "ece": 0.11389948643175861, "mean_confidence": 0.3331357181858688, "confident_error_rate": 0.0, "score_mae": 1.1239995544225962 }, "measuring_hate_speech": { "count": 100, "accuracy": 0.84, "brier": 0.269793616526203, "brier_soft": 0.14445387048134756, "nll": 0.50668914734155, "nll_soft": 0.6627516478286205, "ece": 0.15277715704079395, "mean_confidence": 0.7027064112837865, "confident_error_rate": 0.0, "score_mae": 0.542979770335009 }, "medqa": { "count": 100, "accuracy": 0.61, "brier": 0.5232384116814891, "brier_soft": 0.5232384116814891, "nll": 0.9666950597140125, "nll_soft": 0.9666950597140125, "ece": 0.15310429795340835, "mean_confidence": 0.7327647249737831, "confident_error_rate": 0.02 }, "mnli": { "count": 100, "accuracy": 0.9, "brier": 0.1349148983200619, "brier_soft": 0.1349148983200619, "nll": 0.2432668425098134, "nll_soft": 0.2432668425098134, "ece": 0.049021119706413305, "mean_confidence": 0.906729724583111, "confident_error_rate": 0.02 }, "mrpc": { "count": 100, "accuracy": 0.81, "brier": 0.2675980359243206, "brier_soft": 0.2675980359243206, "nll": 0.399899604734159, "nll_soft": 0.399899604734159, "ece": 0.09618112086460426, "mean_confidence": 0.8472228896042645, "confident_error_rate": 0.01 }, "multirc": { "count": 125, "accuracy": 0.752, "brier": 0.3283689232391426, "brier_soft": 0.3283689232391426, "nll": 0.4942773812599929, "nll_soft": 0.4942773812599929, "ece": 0.11061435237291556, "mean_confidence": 0.8251074372391113, "confident_error_rate": 0.032, "evidence_auc": 0.9989333333333333 }, "openbookqa": { "count": 100, "accuracy": 0.87, "brier": 0.18715713115610605, "brier_soft": 0.18715713115610605, "nll": 0.32547382828638893, "nll_soft": 0.32547382828638893, "ece": 0.07002363675922497, "mean_confidence": 0.910568385075042, "confident_error_rate": 0.01 }, "piqa": { "count": 100, "accuracy": 0.82, "brier": 0.2539820134589331, "brier_soft": 0.2539820134589331, "nll": 0.4068688007847875, "nll_soft": 0.4068688007847875, "ece": 0.08311327451943891, "mean_confidence": 0.8653291206132601, "confident_error_rate": 0.04 }, "policy_application": { "count": 150, "accuracy": 0.9066666666666666, "brier": 0.14813577608521744, "brier_soft": 0.14813577608521744, "nll": 0.23572050653690194, "nll_soft": 0.23572050653690194, "ece": 0.06691158593503817, "mean_confidence": 0.9575142657331253, "confident_error_rate": 0.02666666666666667, "evidence_auc": 1.0 }, "policy_hard": { "count": 300, "accuracy": 0.8433333333333334, "brier": 0.21603088941877208, "brier_soft": 0.21603088941877208, "nll": 0.3671480630817236, "nll_soft": 0.3671480630817236, "ece": 0.07672543440705276, "mean_confidence": 0.909976742390077, "confident_error_rate": 0.04666666666666667, "evidence_auc": 0.9817708333333334 }, "qqp": { "count": 100, "accuracy": 0.91, "brier": 0.1356641944974391, "brier_soft": 0.1356641944974391, "nll": 0.2231311517649759, "nll_soft": 0.2231311517649759, "ece": 0.06575801900138287, "mean_confidence": 0.8946227147846244, "confident_error_rate": 0.02 }, "race": { "count": 125, "accuracy": 0.72, "brier": 0.35765456161571446, "brier_soft": 0.35765456161571446, "nll": 0.6983141916139135, "nll_soft": 0.6983141916139135, "ece": 0.0932914444873349, "mean_confidence": 0.8087819135327284, "confident_error_rate": 0.032, "evidence_auc": 0.9992 }, "rotten_tomatoes": { "count": 100, "accuracy": 0.94, "brier": 0.08124824522675228, "brier_soft": 0.08124824522675228, "nll": 0.12389669533170551, "nll_soft": 0.12389669533170551, "ece": 0.05773620838858384, "mean_confidence": 0.9526650174313278, "confident_error_rate": 0.01 }, "rte": { "count": 100, "accuracy": 0.91, "brier": 0.18311153542500938, "brier_soft": 0.18311153542500938, "nll": 0.3545909140092871, "nll_soft": 0.3545909140092871, "ece": 0.10124014569716058, "mean_confidence": 0.943752464163045, "confident_error_rate": 0.06 }, "scotus": { "count": 100, "accuracy": 0.21, "brier": 0.89769193671844, "brier_soft": 0.89769193671844, "nll": 2.415346637641461, "nll_soft": 2.415346637641461, "ece": 0.07353758881534735, "mean_confidence": 0.24746621955170325, "confident_error_rate": 0.0 }, "shp": { "count": 100, "accuracy": 0.68, "brier": 0.4733767173132663, "brier_soft": 0.10233528576029081, "nll": 0.6665194253969489, "nll_soft": 0.6859362143369068, "ece": 0.14897214933195024, "mean_confidence": 0.546426399891975, "confident_error_rate": 0.0 }, "sms_spam": { "count": 100, "accuracy": 0.99, "brier": 0.01930635801208172, "brier_soft": 0.01930635801208172, "nll": 0.04207345377972822, "nll_soft": 0.04207345377972822, "ece": 0.0067073655477912775, "mean_confidence": 0.9967073655477913, "confident_error_rate": 0.01 }, "snli": { "count": 100, "accuracy": 0.88, "brier": 0.17582400400290324, "brier_soft": 0.17582400400290324, "nll": 0.30302966915953083, "nll_soft": 0.30302966915953083, "ece": 0.0709306654852658, "mean_confidence": 0.9286928509331307, "confident_error_rate": 0.04 }, "social_iqa": { "count": 100, "accuracy": 0.78, "brier": 0.3177565365861535, "brier_soft": 0.3177565365861535, "nll": 0.607449232287739, "nll_soft": 0.607449232287739, "ece": 0.12725246333797624, "mean_confidence": 0.8733738051484262, "confident_error_rate": 0.04 }, "sst2": { "count": 100, "accuracy": 0.96, "brier": 0.07971217148658985, "brier_soft": 0.07971217148658985, "nll": 0.17264823580876176, "nll_soft": 0.17264823580876176, "ece": 0.043530750660772013, "mean_confidence": 0.9733277126761404, "confident_error_rate": 0.03 }, "sst5": { "count": 100, "accuracy": 0.67, "brier": 0.48402215895940337, "brier_soft": 0.48402215895940337, "nll": 0.8604305286260937, "nll_soft": 0.8604305286260937, "ece": 0.16751759519570414, "mean_confidence": 0.5497045612334717, "confident_error_rate": 0.0, "score_mae": 0.4519901699493576 }, "stance_abortion": { "count": 60, "accuracy": 0.6666666666666666, "brier": 0.45936729035760654, "brier_soft": 0.45936729035760654, "nll": 0.705900344737038, "nll_soft": 0.705900344737038, "ece": 0.15772369018233034, "mean_confidence": 0.8077103918863352, "confident_error_rate": 0.03333333333333333 }, "stance_atheism": { "count": 60, "accuracy": 0.7833333333333333, "brier": 0.3328091281497823, "brier_soft": 0.3328091281497823, "nll": 0.5367039070600078, "nll_soft": 0.5367039070600078, "ece": 0.14772804545430168, "mean_confidence": 0.8721167817648096, "confident_error_rate": 0.06666666666666667 }, "stance_feminist": { "count": 60, "accuracy": 0.75, "brier": 0.38284189015448966, "brier_soft": 0.38284189015448966, "nll": 0.6097841885696503, "nll_soft": 0.6097841885696503, "ece": 0.15592787843210057, "mean_confidence": 0.8234037784819485, "confident_error_rate": 0.016666666666666666 }, "stance_hillary": { "count": 60, "accuracy": 0.6666666666666666, "brier": 0.4047196431521728, "brier_soft": 0.4047196431521728, "nll": 0.6834731284888856, "nll_soft": 0.6834731284888856, "ece": 0.1269529390022274, "mean_confidence": 0.7863250004418558, "confident_error_rate": 0.05 }, "stsb": { "count": 100, "accuracy": 0.54, "brier": 0.5614097077774185, "brier_soft": 0.2704198364032701, "nll": 0.9670080291366133, "nll_soft": 0.9841675945745174, "ece": 0.08449101891072369, "mean_confidence": 0.5913493599636942, "confident_error_rate": 0.0, "score_mae": 0.5504846148923596 }, "subjectivity": { "count": 100, "accuracy": 0.94, "brier": 0.10407316114282592, "brier_soft": 0.10407316114282592, "nll": 0.17088805058804918, "nll_soft": 0.17088805058804918, "ece": 0.040859098694625375, "mean_confidence": 0.9664050958509474, "confident_error_rate": 0.04 }, "swag": { "count": 100, "accuracy": 0.77, "brier": 0.3051043063037333, "brier_soft": 0.3051043063037333, "nll": 0.5433260597527682, "nll_soft": 0.5433260597527682, "ece": 0.08123775745211292, "mean_confidence": 0.8281818811177126, "confident_error_rate": 0.03 }, "trec": { "count": 100, "accuracy": 0.97, "brier": 0.06155337791663981, "brier_soft": 0.06155337791663981, "nll": 0.1634662721991823, "nll_soft": 0.1634662721991823, "ece": 0.02569771263554021, "mean_confidence": 0.9807213871260368, "confident_error_rate": 0.03 }, "tweet_hate": { "count": 100, "accuracy": 0.48, "brier": 0.7950030937517019, "brier_soft": 0.7950030937517019, "nll": 1.3997674961537496, "nll_soft": 1.3997674961537496, "ece": 0.40343398490604676, "mean_confidence": 0.8739591757981232, "confident_error_rate": 0.28 }, "tweet_sentiment": { "count": 100, "accuracy": 0.66, "brier": 0.42840573743009913, "brier_soft": 0.42840573743009913, "nll": 0.7082548515314017, "nll_soft": 0.7082548515314017, "ece": 0.1119492420242185, "mean_confidence": 0.6316418997585348, "confident_error_rate": 0.0, "score_mae": 0.3841625395096181 }, "vitaminc": { "count": 100, "accuracy": 0.84, "brier": 0.27409571807544036, "brier_soft": 0.27409571807544036, "nll": 0.546105934849192, "nll_soft": 0.546105934849192, "ece": 0.10942523312978639, "mean_confidence": 0.9069065792619518, "confident_error_rate": 0.09 }, "wic": { "count": 100, "accuracy": 0.71, "brier": 0.36201195057161817, "brier_soft": 0.36201195057161817, "nll": 0.5375649578022699, "nll_soft": 0.5375649578022699, "ece": 0.08803995502979516, "mean_confidence": 0.7847677951555196, "confident_error_rate": 0.03 }, "winogrande": { "count": 100, "accuracy": 0.75, "brier": 0.392578222900395, "brier_soft": 0.392578222900395, "nll": 0.694012713399598, "nll_soft": 0.694012713399598, "ece": 0.1698561596606002, "mean_confidence": 0.8960738105444639, "confident_error_rate": 0.11 }, "yelp": { "count": 100, "accuracy": 0.68, "brier": 0.44095049816750126, "brier_soft": 0.44095049816750126, "nll": 0.7369559085064258, "nll_soft": 0.7369559085064258, "ece": 0.10563042684119633, "mean_confidence": 0.6006492517646783, "confident_error_rate": 0.0, "score_mae": 0.4214742136848633 } }, "reliability": [ { "bin": "0.0-0.1", "count": 0, "confidence": NaN, "accuracy": NaN }, { "bin": "0.1-0.2", "count": 26, "confidence": 0.1786501163201734, "accuracy": 0.23076923076923078 }, { "bin": "0.2-0.3", "count": 138, "confidence": 0.2481680800607115, "accuracy": 0.21014492753623187 }, { "bin": "0.3-0.4", "count": 126, "confidence": 0.3559411667797675, "accuracy": 0.2857142857142857 }, { "bin": "0.4-0.5", "count": 344, "confidence": 0.4660676106444815, "accuracy": 0.502906976744186 }, { "bin": "0.5-0.6", "count": 542, "confidence": 0.5533377851748736, "accuracy": 0.566420664206642 }, { "bin": "0.6-0.7", "count": 418, "confidence": 0.6479330715953778, "accuracy": 0.6004784688995215 }, { "bin": "0.7-0.8", "count": 466, "confidence": 0.7491832278985806, "accuracy": 0.6888412017167382 }, { "bin": "0.8-0.9", "count": 564, "confidence": 0.8562632478879275, "accuracy": 0.7464539007092199 }, { "bin": "0.9-1.0", "count": 3489, "confidence": 0.9810906862120456, "accuracy": 0.942676984809401 } ], "seconds": 67.76366090774536, "model": "runs/assay-4b-v4", "data": "data/v4/dev.jsonl" }