Understanding and shaping the behaviour of Large Language Models (LLMs) is increasingly important as applications become more powerful and more frequently adopted. This paper introduces a machine unlearning method specifically designed for LLMs. We introduce a selective pruning method for LLMs that removes neurons based on their relative importance on a targeted capability compared to overall network performance. This approach is a compute- and data-efficient method for identifying and removing neurons that enable specific behaviours. Our findings reveal that both feed-forward and attention neurons in LLMs are specialized; that is, for specific tasks, certain neurons are more crucial than others.

翻译：理解并塑造大型语言模型（LLMs）的行为正变得越来越重要，因为这些应用日益强大且被频繁采用。本文介绍了一种专门为大型语言模型设计的机器遗忘方法。我们提出了一种针对LLMs的选择性剪枝方法，该方法基于神经元在目标能力上的相对重要性（相对于整体网络性能）来移除神经元。这种方法是一种高效计算与数据的方法，用于识别并移除促成特定行为的神经元。我们的研究结果表明，LLMs中的前馈神经元和注意力神经元均有专长化特征；即，对于特定任务，某些神经元比其他神经元更为关键。