arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2503.04693cs.CL
收藏

UIPE:通过移除与遗忘目标相关的知识来增强 LLM 遗忘效果

UIPE: Enhancing LLM Unlearning by Removing Knowledge Related to Forgetting Targets

Shandong University(山东大学);Beijing University of Posts and Telecommunications(北京邮电大学);Leiden University(莱顿大学)

Wenyu Wang, Mengqi Zhang, Xiaotian Ye, Zhaochun Ren, Zhumin Chen, Pengjie Ren

AI总结:

针对LLM遗忘方法因忽略逻辑相关知识导致模型可重构目标内容的问题,提出UIPE方法通过移除与遗忘目标高度相关的知识来增强遗忘效果,在TOFU基准上显著提升了多种主流方法性能。

摘要:

大型语言模型(LLMs)在海量数据集上训练时不可避免地会获取有害信息。LLM 遗忘旨在消除此类有害信息的影响,同时保持模型的整体性能。以基于梯度上升的方法为代表的现有遗忘方法,主要关注遗忘目标数据,却忽略了逻辑相关知识对遗忘效果的关键影响。本文通过理论与实验分析,首先证明遗忘性能不理想的一个关键原因是模型能利用逻辑相关知识进行推理,从而重构目标内容。为解决此问题,我们提出了参数外推遗忘改进(UIPE),该方法移除与遗忘目标高度相关的知识。实验结果表明,UIPE 显著提升了多种主流 LLM 遗忘方法在 TOFU 基准上的表现。

英文摘要:

Large Language Models (LLMs) inevitably acquire harmful information during training on massive datasets. LLM unlearning aims to eliminate the influence of such harmful information while maintaining the model's overall performance. Existing unlearning methods, represented by gradient ascent-based approaches, primarily focus on forgetting target data while overlooking the crucial impact of logically related knowledge on the effectiveness of unlearning. In this paper, through both theoretical and experimental analyses, we first demonstrate that a key reason for the suboptimal unlearning performance is that models can reconstruct the target content through reasoning with logically related knowledge. To address this issue, we propose Unlearning Improvement via Parameter Extrapolation (UIPE), a method that removes knowledge highly correlated with the forgetting targets. Experimental results show that UIPE significantly enhances the performance of various mainstream LLM unlearning methods on the TOFU benchmark.

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。

↑