揭示微调式遗忘方法实为行为控制而非真正删除知识
Dissecting Fine-Tuning Unlearning in Large Language Models
- 通过激活修补与参数恢复实验，发现方法改变知识检索路径
- 最终层MLP系数主导遗忘效果，非真正知识消除
- 遗忘会波及无关知识，影响模型整体能力
基于微调的遗忘方法广泛用于消除大语言模型中的有害、敏感或受版权保护的信息，同时保持整体能力。然而，这些方法的实际效果尚不明确。本文通过激活修补和参数恢复实验，深入探究其局限性。结果表明，这些方法并未真正擦除模型参数中的问题知识，而是改变了知识检索过程。模型最后一层MLP组件生成的系数是产生看似有效遗忘效果的主要因素，对行为控制起关键作用。此外，行为测试显示，该遗忘机制不可避免地影响模型全局行为，损害无关知识或能力。代码已公开于https://github.com/yihuaihong/Dissecting-FT-Unlearning。
原文摘要 · Abstract (English)
Fine-tuning-based unlearning methods prevail for preventing targeted harmful, sensitive, or copyrighted information within large language models while preserving overall capabilities. However, the true effectiveness of these methods is unclear. In this work, we delve into the limitations of fine-tuning-based unlearning through activation patching and parameter restoration experiments. Our findings reveal that these methods alter the model's knowledge retrieval process, providing further evidence that they do not genuinely erase the problematic knowledge embedded in the model parameters. Instead, the coefficients generated by the MLP components in the model's final layer are the primary contributors to these seemingly positive unlearning effects, playing a crucial role in controlling the model's behaviors. Furthermore, behavioral tests demonstrate that this unlearning mechanism inevitably impacts the global behavior of the models, affecting unrelated knowledge or capabilities. The code is released at https://github.com/yihuaihong/Dissecting-FT-Unlearning.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。