SCI-PRM: tool-aware process reward model for scientific reasoning verification

Arxiv CS.AI (top papers)

Proposes SCI-PRM, a process reward model adapted for scientific reasoning in biology, chemistry, and physics. Incorporates tool awareness to verify domain-specific reasoning steps.

reward-modeling

Arxiv CS.AI (top papers)·Xiangyu ZhaoHengyuan ZhaoYiheng WangWanghan XuYuhao ZhouQinglong CaoZhiwang ZhouLei BaiWenlong ZhangXiao-Ming Wu

7 days ago