一作ACL 2026Poster· Natural Language Processing
Teach a Reward Model to Correct Itself: Reward Guided Adversarial Failure Discovery for Robust Reward Modeling
Pankayaraj Pathmanathan · Furong Huang
摘要
Reward models (RMs) trained from human preferences are central to aligning large language models, yet they often break under distribution shift or targeted perturbations. Existing failure discovery methods rely on prior knowledge of preference attributes and therefore do not scale to new models o...