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

Metric Correlation with Human Judgment on Merlin

0.369Pearson Correlation

RadGraph

Updated 1mo ago

Evaluation Results

Method	Links
RadGraph 2026.04		0.369	0.251	0.341
RaTEScore 2026.04		0.334	0.227	0.303
CT-FineBench 2026.04		0.326	0.306	0.401
BERTScore 2026.04		0.311	0.217	0.293
GREEN 2026.04		0.309	0.179	0.228
BLEU-2 2026.04		0.168	0.089	0.114
ROUGE-L 2026.04		0.124	0.07	0.097