TruthfulQA
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
TruthfulQA 817 samples (full)
27.78MC1 Score
10
TruthfulQA
83.3Accuracy
10
TruthfulQA (test)
82.33HALL Score
10
TruthfulQA
96.81Correlation Score
10
TruthfulQA
49BLEU Score
10
TruthfulQA
12TruthfulQA Delta
10
TruthfulQA multiple-choice
19.9TruthfulQA Delta (Δ)
10
TruthfulQA MC2
77.72Score
10
TruthfulQA
81.6True*Info Score (TQA)
10
TruthfulQA Two options -> Free-form
52.89Benign F1 Score
10
TruthfulQA Two options -> Four options
57Benign F1
10
TruthfulQA Two options
66.24Benign F1 Score
10
TruthfulQA
68.69Truthfulness Avg.@8
10
TruthfulQA (test)
96.34Truthfulness Score Rate (TSR)
10
TruthfulQA Estonian
43.7Accuracy
10
TruthfulQA open-ended
51.2BLEU
10
TruthfulQA
36.91Win Rate
10
TruthfulQA (test)
2.8LCTG Error Rate
10
TruthfulQA
58.76Normalized Accuracy
10
TruthfulQA
0.68AUROC
9
TruthfulQA
19R@1 Accuracy
9
TruthfulQA
61.08Macro-F1
9
TruthfulQA
2.51Wasserstein Distribution Distance
9
TruthfulQA
70.4Faithfulness Divergence Reduction
9
TruthfulQA
0.33Mean Improvement
9