TruthfulQA
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
TruthfulQA
33.7Normalized Probability Mass
3
TruthfulQA DE 6-shot (test)
34.2Normalized Probability Mass
3
TruthfulQA MC
0.888Accuracy
3
TruthfulQA
81.62Accuracy
3
TruthfulQA
40Error Rate (%)
3
TruthfulQA
81Coverage
3
TruthfulQA (test)
1.698BCE
3
TruthfulQA o=5 (f_clean)
55.5Accuracy (Exact)
3
TruthfulQA
27.7MC1
3
TruthfulQA-MC Indonesian
36.6Accuracy
3
TruthfulQA
99.9Quality
2
TruthfulQA
99.5Quality Score
2
TruthfulQA
0Cost (cents)
2
TruthfulQA
99.6Quality (%)
2
TruthfulQA
0.998Sel-AUC
2
TruthfulQA
100Accuracy under Attack
2
TruthfulQA Freeform
63.94Benign F1
2
TruthfulQA FourOptions
80.59Benign F1 Score
2
TruthfulQA TwoOptions
79.81Benign F1 Score
2
TruthfulQA (200 samples)
0.37Log-likelihood Score
2
TruthfulQA
92.2Accuracy
2
TruthfulQA
12Baseline Wins
2
TruthfulQA DE
34.8Normalized Probability Mass
2
TruthfulQA DE
17.4Norm. Prob. Mass
2
TruthfulQA 6-shot
0.346Normalized Probability Mass
2