dyyyyyyyy/r2e-gym-subset-filtered
收藏官方服务：
资源简介：
--- license: apache-2.0 --- This dataset is derived from `r2e-gym-subset`. It keeps only samples whose gold patch verifies correctly in all 16 runs. This is used to make the verify reward score more accurate during RL training, for stable verification.
提供机构：
dyyyyyyyy
