← 研踪 PaperTrailACL Anthology · ACL 2026:acl:2026.acl-long.418 · 记录 #1
Teach a Reward Model to Correct Itself: Reward Guided Adversarial Failure Discovery for Robust Reward Modeling
来源日期（仅到月份） 2026-07
来源记录日期 2026-07
雷达观测 2026-11-01 22:02
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 or data. We introduce a preference distribution agnostic procedure that uses the reward model itself to guide controlled decoding toward mis specified responses while preserving the underlying preference class. Building on this discovery mechanism, we propose REFORM, a self improving RM framework that (i) searches for class consistent but reward inconsistent variants and (ii) fine tunes the RM on a small, targeted augmentation of these failures. On Anthropic Helpful Harmless and PKU Beavertails, REFORM consistently improves robustness without degrading in distribution reward quality across different models (e.g., Mistral-7B and Qwen-14B), with an average improvement of 35%–45%.Further, across Best of N sampling, PPO, and DPO, REFORM preserves downstream generation quality and reduces spurious correlations. Our results show that RMs can serve as their own adversary to expose and fix blind spots, yielding robust alignment without manual attribute priors or large scale relabeling.
论文发现出处
这些页面提供论文线索，不代表同行评审或质量认证。
观测到的版本
平台记录号代表本站观测到的元数据变化，不是作者修订号。来源日期仅到月份或年份时，按该月或该年首日存档。
记录 #1 · 2026-07 · 最新观测版本
Teach a Reward Model to Correct Itself: Reward Guided Adversarial Failure Discovery for Robust Reward Modeling
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 or data. We introduce a preference distribution agnostic procedure that uses the reward model itself to guide controlled decoding toward mis specified responses while preserving the underlying preference class. Building on this discovery mechanism, we propose REFORM, a self improving RM framework that (i) searches for class consistent but reward inconsistent variants and (ii) fine tunes the RM on a small, targeted augmentation of these failures. On Anthropic Helpful Harmless and PKU Beavertails, REFORM consistently improves robustness without degrading in distribution reward quality across different models (e.g., Mistral-7B and Qwen-14B), with an average improvement of 35%–45%.Further, across Best of N sampling, PPO, and DPO, REFORM preserves downstream generation quality and reduces spurious correlations. Our results show that RMs can serve as their own adversary to expose and fix blind spots, yielding robust alignment without manual attribute priors or large scale relabeling.
查看此版本解读 →查看此版本原文 ↗