DUET通过高效教师模型实现轻量级且安全的LLM知识遗忘。
DUET: Distilled LLM Unlearning from an Efficiently Contextualized Teacher
- 用提示引导的教师模型指导学生模型，模仿其拒绝有害知识的行为。
- 在遗忘率和通用知识保留上优于现有方法，数据效率提升数个数量级。
- 适合需要安全、高效知识清除的可信AI场景。
大语言模型（LLM）的知识遗忘技术旨在移除模型中不期望的知识，而无需从头训练，这对构建可信AI至关重要。现有方法存在显著局限：基于微调的遗忘方法计算开销大，易引发灾难性遗忘；基于上下文的遗忘方法虽轻量且精准，但易受提示删除或逆向工程攻击影响。为此，我们提出一种新的基于蒸馏的遗忘方法——高效教师驱动的蒸馏遗忘（DUET），该方法让学生模型模仿一个由提示控制的教师模型，该教师模型能有效拒绝生成有害知识，同时保留通用领域知识。在多个基准测试上的广泛评估表明，与现有方法相比，DUET在遗忘效果和知识保留性能上均更优，且数据效率达到当前最优方法的数个数量级。
原文摘要 · Abstract (English)
LLM unlearning is a technique to remove the impacts of undesirable knowledge from the model without retraining from scratch, which is indispensable towards trustworthy AI. Existing unlearning methods face significant limitations: conventional tuning-based unlearning is computationally heavy and prone to catastrophic forgetting. In contrast, in-contextualized unlearning is lightweight for precise unlearning but vulnerable to prompt removal or reverse engineering attacks. In response, we propose Distilled Unlearning from an Efficient Teacher (DUET), a novel distillation-based unlearning method that combines the merits of these two lines of work. It learns a student model to imitate the behavior of a prompt-steered teacher that effectively refuses undesirable knowledge generation while preserving general domain knowledge. Extensive evaluations on existing benchmarks with our enriched evaluation protocols demonstrate that DUET achieves higher performance in both forgetting and utility preservation, while being orders of magnitude more data-efficient than state-of-the-art unlearning methods.
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。