HaluEval
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
HaluEval (test)
98.55AUC-ROC
176
HaluEval
1AUROC
135
HaluEval
100Accuracy (ACC)
51
HaluEvalQA
89ROC-AUC
39
HaluEval Sum (500 items)
63MC1 Score
30
HaluEval QA (500 items)
86.4MC1 Score
30
HaluEval Adversarial
100nAOPC
28
HaluEval (in-distribution)
99.1Accuracy
25
HaluEval Dialogue latest (test)
84.88Accuracy
22
HaluEval QA
99.5Accuracy
17
HaluEval (test)
93.89QA Accuracy
16
HaluEval Gemini outputs (test)
0.571AUROC
15
HaluEval GPT outputs (test)
0.582AUROC
15
HaluEval Llama outputs (test)
0.704AUROC
15
HaluEval
72.2Dialogue Score
15
HaluEval
68.7Accuracy (Response)
14
HaluEval QA
45.4Accuracy
14
HaluEval held-out 50% (test)
69.9AUROC
12
HaluEval DA
77.1F1 Score
12
HaluEval
68EM
12
HaluEval Sum
65.9F1 Score
12
HaluEval
72.66F1 Score
11
HaluEval
0.78Kendall's Tau
11
HaluEval (test)
50.33HALL Rate
10
HaluEval
21.3HaluEval Delta
10