REFUTE-knowledge-v3
收藏官方服务：
资源简介：
REFUTE knowledge-recall subset (scientific critique benchmark). 4-way MCQ grounded in recent science paper summaries. Target column answer is A-D. Preprint DOI: https://doi.org/10.5281/zenodo.21542238 Product: https://bgpt.pro/refute
创建时间：
2026-07-25
