在悬崖边转向还是偏离？重新思考推断时间干预中的特异性和鲁棒性
Steering Safely or Off a Cliff? Rethinking Specificity and Robustness in Inference-Time Interventions
AI总结：
本文提出模型转向中特异性和鲁棒性的评估框架，揭示转向方法在提升效率的同时可能牺牲安全性。
摘要：
在推断时间对隐藏表示进行干预的模型转向技术，已成为一种轻量级的替代方法，用于精确控制大语言模型。尽管转向的有效性已被广泛研究，但评估干预是否仅改变预期属性的评估仍有限，尤其是在与目标属性相关的意外行为变化方面。我们称之为特异性。我们提出一个框架，区分特异性的三个维度：通用性（保持流畅性和无关能力）、控制性（保持相关控制属性）和鲁棒性（在分布偏移下保持控制属性）。我们研究了两个安全关键的应用场景：将模型转向以减少过度拒绝和忠实性幻觉，并展示转向在实现高效率和维持通用性和控制性特异性方面表现良好，但始终无法保持鲁棒性特异性。例如，在过度拒绝转向的情况下，所有转向方法都能减少过度拒绝而不损害通用能力和拒绝有害查询的能力；然而，它们显著增加了对劫持的易感性。我们的工作提供了模型转向中特异性的首次系统评估，表明标准的有效性和特异性检查不足，因为没有鲁棒性评估，转向方法可能看起来可靠，即使它们损害了模型安全性。
英文摘要：
Model steering, which involves intervening on hidden representations at inference time, has emerged as a lightweight alternative to finetuning for precisely controlling large language models. While steering efficacy has been widely studied, evaluations of whether interventions alter only the intended property remain limited, especially with respect to unintended changes in behaviors related to the target property. We call this notion specificity. We propose a framework that distinguishes three dimensions of specificity: general (preserving fluency and unrelated abilities), control (preserving related control properties), and robustness (preserving control properties under distribution shifts). We study two safety-critical use cases: steering models to reduce overrefusal and faithfulness hallucinations, and show that while steering achieves high efficacy and largely maintains general and control specificity, it consistently fails to preserve robustness specificity. In the case of overrefusal steering, for example, all steering methods reduce overrefusal without harming general abilities and refusal on harmful queries; however, they substantially increase vulnerability to jailbreaks. Our work provides the first systematic evaluation of specificity in model steering, showing that standard efficacy and specificity checks are insufficient, because without robustness evaluation, steering methods may appear reliable even when they compromise model safety.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
