Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

TruthfulQA

Benchmarks

Task NameDataset NameSOTA ResultTrend
Hallucination DetectionTruthfulQA
AUC (ROC)0.9417
182
Question AnsweringTruthfulQA
Accuracy86.6
164
FactualityTruthfulQA
Accuracy91.5
145
Hallucination DetectionTruthfulQA (test)
AUC-ROC89.5
112
Truthfulness EvaluationTruthfulQA
Accuracy75
108
Factuality EvaluationTruthfulQA
MC294.3
103
TruthfulnessTruthfulQA
Truthfulness Accuracy97.55
98
Hallucination DetectionTruthfulQA
AUROC0.8851
91
Multiple-ChoiceTruthfulQA
MC1 Accuracy58.5
83
TruthfulnessTruthfulQA
Truthfulness Accuracy72.36
78
Question AnsweringTruthfulQA
Accuracy86.64
73
Selective GenerationTruthfulQA
ROC-AUC0.744
66
Truthful Question AnsweringTruthfulQA MC2
MC2 Accuracy56.46
66
Question AnsweringTruthfulQA
TruthfulQA Score63
61
Truthfulness EvaluationTruthfulQA
T·I Score84.7
59
Question AnsweringTruthfulQA MC1
MC1 Accuracy88.8
54
Machine-Generated Text DetectionTruthfulQA
TPR@FPR-1% (ChatGLM)98.38
54
Question AnsweringTruthfulQA
Performance Score81.1
52
Open-ended GenerationTruthfulQA
BLEURT Score70.13
48
Predicting answer correctnessTruthfulQA
AUROC0.7272
48
Truthful and Informative GenerationTruthfulQA (test)
True*Info (%)84.7
44
Question AnsweringTruthfulQA
MC268.25
43
Generation correctness predictionTruthfulQA (test)
AURC62.69
42
Uncertainty EstimationTruthfulQA
PRR26.8
42
HallucinationTruthfulQA
Score75.76
42
Showing 25 of 255 rows
...