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

Factual Consistency Detection on TofuEval-MeetingBank Span level leakage-clean protocol (test)

0.693AUC

GASP-threshold (default)

0.507880.555940.6040.65206Jul 5, 2026
Updated 18d ago

Evaluation Results

MethodLinks
2026.07
0.693
2026.07
0.693
2026.07
0.693
2026.07
0.68
2026.07
0.675
2026.07
0.672
2026.07
0.656
2026.07
0.642
2026.07
0.631
2026.07
0.555
2026.07
0.535
2026.07
0.524
2026.07
0.516
2026.07
0.515
2026.07
0.515