跳到正文
arXiv · AI· Daoan Zhang, Qinbo Bai, Dehai Min, Xingyu Mu, Ziyi Chen, Jianing Yin·· 9 天前AI 评分42

Veritas: A Harness for Automatically Generating Agent Task Verifiers

Veritas: A Verifier Generation Harness for Agent Tasks

AI 导读

Self-improving agent systems rely on high-quality verifiers to evaluate outcomes。

来源:arXiv · AI · arxiv.org