{ "overall": { "questions": 908, "answered": 908, "errors": 0, "correct": 783, "accuracyAllQuestions": 0.8623348017621145, "brierAnswered": 0.19616500539754675, "nllAnswered": 0.374025080065555, "ece10Answered": 0.04207503699664424, "confidentWrongAt09": 26, "coverageByFixedThreshold": [ { "threshold": 0.5, "accepted": 889, "coverage": 0.9790748898678414, "errorRate": 0.12373453318335208 }, { "threshold": 0.7, "accepted": 819, "coverage": 0.9019823788546255, "errorRate": 0.08424908424908426 }, { "threshold": 0.8, "accepted": 767, "coverage": 0.8447136563876652, "errorRate": 0.06649282920469361 }, { "threshold": 0.9, "accepted": 676, "coverage": 0.7444933920704846, "errorRate": 0.038461538461538464 }, { "threshold": 0.95, "accepted": 567, "coverage": 0.6244493392070485, "errorRate": 0.012345679012345678 }, { "threshold": 0.99, "accepted": 0, "coverage": 0.0, "errorRate": null } ], "coverageInterpretation": "Observed on this dataset; not a future error guarantee. Thresholds not tuned here." }, "byType": { "choice": { "questions": 492, "answered": 492, "errors": 0, "correct": 400, "accuracyAllQuestions": 0.8130081300813008, "brierAnswered": 0.2615263089500927, "nllAnswered": 0.5247360934156637, "ece10Answered": 0.06533461933177653, "confidentWrongAt09": 20, "coverageByFixedThreshold": [ { "threshold": 0.5, "accepted": 473, "coverage": 0.9613821138211383, "errorRate": 0.16279069767441862 }, { "threshold": 0.7, "accepted": 424, "coverage": 0.8617886178861789, "errorRate": 0.10377358490566038 }, { "threshold": 0.8, "accepted": 386, "coverage": 0.7845528455284553, "errorRate": 0.08031088082901554 }, { "threshold": 0.9, "accepted": 331, "coverage": 0.6727642276422764, "errorRate": 0.06042296072507553 }, { "threshold": 0.95, "accepted": 267, "coverage": 0.5426829268292683, "errorRate": 0.026217228464419477 }, { "threshold": 0.99, "accepted": 0, "coverage": 0.0, "errorRate": null } ], "coverageInterpretation": "Observed on this dataset; not a future error guarantee. Thresholds not tuned here." }, "noul": { "questions": 328, "answered": 328, "errors": 0, "correct": 298, "accuracyAllQuestions": 0.9085365853658537, "brierAnswered": 0.13871514597249474, "nllAnswered": 0.22139737153892355, "ece10Answered": 0.04994271263817423, "confidentWrongAt09": 5, "coverageByFixedThreshold": [ { "threshold": 0.5, "accepted": 328, "coverage": 1.0, "errorRate": 0.09146341463414634 }, { "threshold": 0.7, "accepted": 309, "coverage": 0.9420731707317073, "errorRate": 0.07766990291262135 }, { "threshold": 0.8, "accepted": 296, "coverage": 0.9024390243902439, "errorRate": 0.06418918918918919 }, { "threshold": 0.9, "accepted": 261, "coverage": 0.7957317073170732, "errorRate": 0.019157088122605363 }, { "threshold": 0.95, "accepted": 220, "coverage": 0.6707317073170732, "errorRate": 0.0 }, { "threshold": 0.99, "accepted": 0, "coverage": 0.0, "errorRate": null } ], "coverageInterpretation": "Observed on this dataset; not a future error guarantee. Thresholds not tuned here." }, "score": { "questions": 88, "answered": 88, "errors": 0, "correct": 85, "accuracyAllQuestions": 0.9659090909090909, "brierAnswered": 0.04486719339259727, "nllAnswered": 0.10029860084375529, "ece10Answered": 0.03799778408573792, "confidentWrongAt09": 1, "coverageByFixedThreshold": [ { "threshold": 0.5, "accepted": 88, "coverage": 1.0, "errorRate": 0.03409090909090909 }, { "threshold": 0.7, "accepted": 86, "coverage": 0.9772727272727273, "errorRate": 0.011627906976744186 }, { "threshold": 0.8, "accepted": 85, "coverage": 0.9659090909090909, "errorRate": 0.011764705882352941 }, { "threshold": 0.9, "accepted": 84, "coverage": 0.9545454545454546, "errorRate": 0.011904761904761904 }, { "threshold": 0.95, "accepted": 80, "coverage": 0.9090909090909091, "errorRate": 0.0 }, { "threshold": 0.99, "accepted": 0, "coverage": 0.0, "errorRate": null } ], "coverageInterpretation": "Observed on this dataset; not a future error guarantee. Thresholds not tuned here." } }, "bySource": { "authored-russian-typed": { "questions": 144, "answered": 144, "errors": 0, "correct": 144, "accuracyAllQuestions": 1.0, "brierAnswered": 0.0009998268084272204, "nllAnswered": 0.024466212500916604, "ece10Answered": 0.024139216454452272, "confidentWrongAt09": 0, "coverageByFixedThreshold": [ { "threshold": 0.5, "accepted": 144, "coverage": 1.0, "errorRate": 0.0 }, { "threshold": 0.7, "accepted": 144, "coverage": 1.0, "errorRate": 0.0 }, { "threshold": 0.8, "accepted": 144, "coverage": 1.0, "errorRate": 0.0 }, { "threshold": 0.9, "accepted": 144, "coverage": 1.0, "errorRate": 0.0 }, { "threshold": 0.95, "accepted": 144, "coverage": 1.0, "errorRate": 0.0 }, { "threshold": 0.99, "accepted": 0, "coverage": 0.0, "errorRate": null } ], "coverageInterpretation": "Observed on this dataset; not a future error guarantee. Thresholds not tuned here." }, "composition_holdout": { "questions": 96, "answered": 96, "errors": 0, "correct": 84, "accuracyAllQuestions": 0.875, "brierAnswered": 0.18821988619559113, "nllAnswered": 0.31506340814867745, "ece10Answered": 0.08715274769631735, "confidentWrongAt09": 9, "coverageByFixedThreshold": [ { "threshold": 0.5, "accepted": 96, "coverage": 1.0, "errorRate": 0.125 }, { "threshold": 0.7, "accepted": 92, "coverage": 0.9583333333333334, "errorRate": 0.09782608695652174 }, { "threshold": 0.8, "accepted": 92, "coverage": 0.9583333333333334, "errorRate": 0.09782608695652174 }, { "threshold": 0.9, "accepted": 92, "coverage": 0.9583333333333334, "errorRate": 0.09782608695652174 }, { "threshold": 0.95, "accepted": 86, "coverage": 0.8958333333333334, "errorRate": 0.03488372093023256 }, { "threshold": 0.99, "accepted": 0, "coverage": 0.0, "errorRate": null } ], "coverageInterpretation": "Observed on this dataset; not a future error guarantee. Thresholds not tuned here." }, "emotion": { "questions": 116, "answered": 116, "errors": 0, "correct": 69, "accuracyAllQuestions": 0.5948275862068966, "brierAnswered": 0.5245559470753833, "nllAnswered": 1.0208149336604613, "ece10Answered": 0.12016921458494689, "confidentWrongAt09": 2, "coverageByFixedThreshold": [ { "threshold": 0.5, "accepted": 101, "coverage": 0.8706896551724138, "errorRate": 0.3564356435643564 }, { "threshold": 0.7, "accepted": 70, "coverage": 0.603448275862069, "errorRate": 0.22857142857142856 }, { "threshold": 0.8, "accepted": 44, "coverage": 0.3793103448275862, "errorRate": 0.18181818181818182 }, { "threshold": 0.9, "accepted": 8, "coverage": 0.06896551724137931, "errorRate": 0.25 }, { "threshold": 0.95, "accepted": 0, "coverage": 0.0, "errorRate": null }, { "threshold": 0.99, "accepted": 0, "coverage": 0.0, "errorRate": null } ], "coverageInterpretation": "Observed on this dataset; not a future error guarantee. Thresholds not tuned here." }, "legacy_holdout": { "questions": 80, "answered": 80, "errors": 0, "correct": 77, "accuracyAllQuestions": 0.9625, "brierAnswered": 0.04910351545555701, "nllAnswered": 0.10416551178163372, "ece10Answered": 0.03574970703743059, "confidentWrongAt09": 1, "coverageByFixedThreshold": [ { "threshold": 0.5, "accepted": 80, "coverage": 1.0, "errorRate": 0.0375 }, { "threshold": 0.7, "accepted": 78, "coverage": 0.975, "errorRate": 0.01282051282051282 }, { "threshold": 0.8, "accepted": 77, "coverage": 0.9625, "errorRate": 0.012987012987012988 }, { "threshold": 0.9, "accepted": 76, "coverage": 0.95, "errorRate": 0.013157894736842105 }, { "threshold": 0.95, "accepted": 72, "coverage": 0.9, "errorRate": 0.0 }, { "threshold": 0.99, "accepted": 0, "coverage": 0.0, "errorRate": null } ], "coverageInterpretation": "Observed on this dataset; not a future error guarantee. Thresholds not tuned here." }, "mmlu": { "questions": 116, "answered": 116, "errors": 0, "correct": 86, "accuracyAllQuestions": 0.7413793103448276, "brierAnswered": 0.39416899029285185, "nllAnswered": 0.8394104030580065, "ece10Answered": 0.12528513793721022, "confidentWrongAt09": 9, "coverageByFixedThreshold": [ { "threshold": 0.5, "accepted": 112, "coverage": 0.9655172413793104, "errorRate": 0.23214285714285715 }, { "threshold": 0.7, "accepted": 100, "coverage": 0.8620689655172413, "errorRate": 0.18 }, { "threshold": 0.8, "accepted": 91, "coverage": 0.7844827586206896, "errorRate": 0.14285714285714285 }, { "threshold": 0.9, "accepted": 77, "coverage": 0.6637931034482759, "errorRate": 0.11688311688311688 }, { "threshold": 0.95, "accepted": 42, "coverage": 0.3620689655172414, "errorRate": 0.09523809523809523 }, { "threshold": 0.99, "accepted": 0, "coverage": 0.0, "errorRate": null } ], "coverageInterpretation": "Observed on this dataset; not a future error guarantee. Thresholds not tuned here." }, "paws": { "questions": 80, "answered": 80, "errors": 0, "correct": 68, "accuracyAllQuestions": 0.85, "brierAnswered": 0.20695251414059662, "nllAnswered": 0.3140944708626177, "ece10Answered": 0.08749262200436604, "confidentWrongAt09": 2, "coverageByFixedThreshold": [ { "threshold": 0.5, "accepted": 80, "coverage": 1.0, "errorRate": 0.15 }, { "threshold": 0.7, "accepted": 75, "coverage": 0.9375, "errorRate": 0.12 }, { "threshold": 0.8, "accepted": 73, "coverage": 0.9125, "errorRate": 0.0958904109589041 }, { "threshold": 0.9, "accepted": 65, "coverage": 0.8125, "errorRate": 0.03076923076923077 }, { "threshold": 0.95, "accepted": 58, "coverage": 0.725, "errorRate": 0.0 }, { "threshold": 0.99, "accepted": 0, "coverage": 0.0, "errorRate": null } ], "coverageInterpretation": "Observed on this dataset; not a future error guarantee. Thresholds not tuned here." }, "qnli": { "questions": 80, "answered": 80, "errors": 0, "correct": 74, "accuracyAllQuestions": 0.925, "brierAnswered": 0.1185264487594759, "nllAnswered": 0.19157792439378024, "ece10Answered": 0.07130364771118254, "confidentWrongAt09": 1, "coverageByFixedThreshold": [ { "threshold": 0.5, "accepted": 80, "coverage": 1.0, "errorRate": 0.075 }, { "threshold": 0.7, "accepted": 79, "coverage": 0.9875, "errorRate": 0.0759493670886076 }, { "threshold": 0.8, "accepted": 77, "coverage": 0.9625, "errorRate": 0.06493506493506493 }, { "threshold": 0.9, "accepted": 71, "coverage": 0.8875, "errorRate": 0.014084507042253521 }, { "threshold": 0.95, "accepted": 60, "coverage": 0.75, "errorRate": 0.0 }, { "threshold": 0.99, "accepted": 0, "coverage": 0.0, "errorRate": null } ], "coverageInterpretation": "Observed on this dataset; not a future error guarantee. Thresholds not tuned here." }, "sciq": { "questions": 116, "answered": 116, "errors": 0, "correct": 113, "accuracyAllQuestions": 0.9741379310344828, "brierAnswered": 0.03409531064541125, "nllAnswered": 0.09114362656929587, "ece10Answered": 0.05454464085684041, "confidentWrongAt09": 0, "coverageByFixedThreshold": [ { "threshold": 0.5, "accepted": 116, "coverage": 1.0, "errorRate": 0.02586206896551724 }, { "threshold": 0.7, "accepted": 114, "coverage": 0.9827586206896551, "errorRate": 0.008771929824561403 }, { "threshold": 0.8, "accepted": 111, "coverage": 0.9568965517241379, "errorRate": 0.009009009009009009 }, { "threshold": 0.9, "accepted": 106, "coverage": 0.9137931034482759, "errorRate": 0.0 }, { "threshold": 0.95, "accepted": 91, "coverage": 0.7844827586206896, "errorRate": 0.0 }, { "threshold": 0.99, "accepted": 0, "coverage": 0.0, "errorRate": null } ], "coverageInterpretation": "Observed on this dataset; not a future error guarantee. Thresholds not tuned here." }, "tweet_offensive": { "questions": 80, "answered": 80, "errors": 0, "correct": 68, "accuracyAllQuestions": 0.85, "brierAnswered": 0.24263742159686036, "nllAnswered": 0.38374648265869776, "ece10Answered": 0.05576197181938556, "confidentWrongAt09": 2, "coverageByFixedThreshold": [ { "threshold": 0.5, "accepted": 80, "coverage": 1.0, "errorRate": 0.15 }, { "threshold": 0.7, "accepted": 67, "coverage": 0.8375, "errorRate": 0.13432835820895522 }, { "threshold": 0.8, "accepted": 58, "coverage": 0.725, "errorRate": 0.1206896551724138 }, { "threshold": 0.9, "accepted": 37, "coverage": 0.4625, "errorRate": 0.05405405405405406 }, { "threshold": 0.95, "accepted": 14, "coverage": 0.175, "errorRate": 0.0 }, { "threshold": 0.99, "accepted": 0, "coverage": 0.0, "errorRate": null } ], "coverageInterpretation": "Observed on this dataset; not a future error guarantee. Thresholds not tuned here." } }, "provenance": { "revision": "3ef077cd7cc3c97d696c18150358ab4de3a6884e", "codeZipSha256": "b3b7808d7c2cbe2ce26d4b31ee0dde2b3e894c6776537931b61cd173ef526f8e", "execution": "reference BF16 backbone, FP32 shared adapter unmerged", "dataSha256": "9266d6ea1ce83c9755a60cf48c7469b0fa3264c229e96f4142d683bada8fef91", "gpu": "NVIDIA A10G", "torch": "2.14.0+cu130", "questionIsolation": false }, "servedModels": [ "fev-4b" ], "records": 812, "startedUtc": null, "wallLatencyMs": { "median": 86.00620499998968, "p95": 102.34360900000183 }, "latencyMeaning": "Sequential wall time including client overhead/cold calls; not model-only time or throughput" }