推理模型遗忘：遗忘痕迹，而不仅仅是答案，同时保留推理技能
Reasoning Model Unlearning: Forgetting Traces, Not Just Answers, While Preserving Reasoning Skills
AI总结：
本文首次系统研究了大型推理模型的机器遗忘问题，发现传统算法无法擦除CoT中的敏感信息，并提出$R^2MU$方法在保留推理能力的同时有效抑制敏感推理痕迹及最终答案生成。
摘要：
大型推理模型（LRMs）的最新进展通过测试时计算实现了强大的思维链（CoT）生成。虽然这些多步推理能力代表了语言模型性能的一个重大里程碑，但它们也引入了新的安全风险。在这项工作中，我们提出了首个系统性研究，以重新审视在LRMs背景下的机器遗忘问题。机器遗忘是指在不需要完全重新训练的情况下，从已训练的模型中移除敏感、有害或不需要的数据或知识的影响的过程。我们表明，最初为非推理模型设计的传统遗忘算法不适用于LRMs。特别是，即使最终答案被成功擦除，敏感信息通常仍保留在中间推理步骤中，即CoT轨迹中。为了应对这一挑战，我们扩展了传统遗忘方法，并提出了用于遗忘的推理感知表示误导（$R^2MU$），这是一种新颖的方法，能有效抑制敏感推理痕迹并防止生成相关的最终答案，同时保留模型的推理能力。我们的实验表明，$R^2MU$显著减少了推理痕迹中的敏感信息泄露，并在DeepSeek-R1-Distill-LLaMA-8B和DeepSeek-R1-Distill-Qwen-14B等最先进模型上评估时，在安全性和推理基准测试中均取得了强劲的性能。
英文摘要：
Recent advances in large reasoning models (LRMs) have enabled strong chain-of-thought (CoT) generation through test-time computation. While these multi-step reasoning capabilities represent a major milestone in language model performance, they also introduce new safety risks. In this work, we present the first systematic study to revisit the problem of machine unlearning in the context of LRMs. Machine unlearning refers to the process of removing the influence of sensitive, harmful, or undesired data or knowledge from a trained model without full retraining. We show that conventional unlearning algorithms, originally designed for non-reasoning models, are inadequate for LRMs. In particular, even when final answers are successfully erased, sensitive information often persists within the intermediate reasoning steps, i.e., CoT trajectories. To address this challenge, we extend conventional unlearning and propose Reasoning-aware Representation Misdirection for Unlearning ($R^2MU$), a novel method that effectively suppresses sensitive reasoning traces and prevents the generation of associated final answers, while preserving the model's reasoning ability. Our experiments demonstrate that $R^2MU$ significantly reduces sensitive information leakage within reasoning traces and achieves strong performance across both safety and reasoning benchmarks, evaluated on state-of-the-art models such as DeepSeek-R1-Distill-LLaMA-8B and DeepSeek-R1-Distill-Qwen-14B.
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。
