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

Token-level Hallucination Detection on AIME 2024 (AUROC/AUPRC)

87.39AUROC

TOKENHD-8B

63.93870.026576.11582.2035May 12, 2026
Updated 21d ago

Evaluation Results

MethodLinks
2026.05
87.3973.59
2026.05
78.859.08
2026.05
72.0644.26
2026.05
69.0449.19
64.8445.62