opv-process-verifier

【效率与协作】Verify long reasoning chains by combining outcome and process verification through active learning. OPV achieves 83.1 F1 on verification—crucial when both final answers and reasoning paths must be validated with minimal annotation.

独立 GitHub 快照 · 自动分类 · 以原作者文档为准