CRJudgeBench: Can AI Detect Plausible but Invalid Code Reviews?
Results show that even state-of-the-art general-purpose LLMs struggle to identify untrustworthy comments, while iterative action-level learning substantially improves the accuracy of repository-grounded trustworthiness judgments.
Yue Pan, Jia-Wei Li, Zi-Yuan Zhang et al.
· 0 citations