TruthfulQA
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
TruthfulQA n=100
33MC1 Score
6
TruthfulQA
5.94Best Probe Selectivity
6
TruthfulQA (Memorization)
46.06ROUGE-L
6
TruthfulQA (in-domain)
0.924AUC
6
TruthfulQA MC2
41.9Accuracy
6
TruthfulQA MC1
26.3Accuracy
6
TruthfulQA (double)
79.3Accuracy
6
TruthfulQA (single)
78.1Accuracy
6
TruthfulQA
76.3TA Rate
6
TruthfulQA
67.25Accuracy
5
TruthfulQA
99.7Sel-AUC
5
TruthfulQA Dr (remaining subset)
93BERTScore
5
TruthfulQA public generation
99.4Coverage
5
TruthfulQA
51.6Utility
5
TruthfulQA
81.88TruthfulQA Score
5
TruthfulQA
62.3TruthfulQA Score
5
TruthfulQA
0.61AUROC (phi C75)
5
TruthfulQA
92.8Accuracy
5
TruthfulQA gen
42BLEU Acc
5
TruthfulQA 2
0.41BLEU
5
TruthfulQA 1
25BLEU
5
TruthfulQA (80% test)
0.689AUROC (T=0)
5
TruthfulQA
82.7Macro F1
4
TruthfulQA
39.41Accuracy
4
TruthfulQA
54.8Accuracy
4