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

Truthfulness Evaluation on TruthfulQA (Accuracy)

65.75Accuracy

LoPT-GRPO

34.8142.842550.87558.9075Apr 21, 2026Apr 27, 2026May 3, 2026May 9, 2026May 15, 2026May 21, 2026May 28, 2026
Updated 5d ago

Evaluation Results

MethodLinks
2026.05
65.75
2026.05
65.73
2026.05
65.54
2026.04
62.51
2026.04
62.35
2026.04
56.4
2026.05
52
2026.05
50
2026.04
49.81
2026.05
49.5
2026.05
49.5
2026.04
48.07
2026.04
48.06
2026.05
46.5
2026.05
46.5
2026.05
46
2026.05
45.5
2026.05
45.5
2026.05
44
2026.05
36