Metric Correlation with Human Judgment on CT-RATE benchmark leaderboard · SOTA2 Research

Metric Correlation with Human Judgment on CT-RATE

0.622Pearson CorrelationCT-FineBench

Evaluation Results

Method	Links
CT-FineBench2026.04		0.622	0.378	0.49
RaTEScore2026.04		0.521	0.32	0.434
BLEU-22026.04		0.495	0.353	0.479
BERTScore2026.04		0.349	0.262	0.342
ROUGE-L2026.04		0.17	0.115	0.158
RadGraph2026.04		0.163	0.137	0.171
GREEN2026.04		0.111	0.063	0.088