MedRECT: A Bilingual Medical Reasoning Benchmark for Error Correction in Clinical Texts
LoRA fine-tuning yields higher sentence extraction accuracy and higher point estimates on all three reference-based correction similarity metrics in both languages, and several leading general-purpose reasoning models outperform all three evaluated medical-domain models on these two subtasks.
Naoto Iwase, Hiroki Okuyama, Junichiro Iwasawa
· 6 citations
· ⚡1