一心多用：深入探究大型语言模型中语言无关的知识神经元
One Mind, Many Tongues: A Deep Dive into Language-Agnostic Knowledge Neurons in Large Language Models
AI总结：
针对大型语言模型中知识存储机制的神秘性及现有语言无关知识神经元研究的局限，该研究构建RML-LAMA基准并提出MATRICE方法，以准确地定位语言无关知识神经元，并探究其在跨语言知识任务中的作用。
摘要：
大型语言模型（LLMs）通过在大规模语料库上的自监督预训练学习了大量事实知识。同时，LLMs也展现出卓越的多语言能力，能以多种语言表达所学知识。然而，LLMs中的知识存储机制仍很神秘。部分研究者试图从知识神经元视角揭开LLMs事实知识的面纱，随后发现了语言无关知识神经元——以跨越语言障碍的形式存储事实知识。但这一初步发现存在两个局限：1）定位结果高度不确定。现有研究仅用基于提示的探针定位每个事实的知识神经元，而LLMs无法为语义等价查询提供一致答案，导致定位结果不准确且不确定性高；2）缺乏更多语言分析。该研究仅在英、中数据上分析语言无关知识神经元，未探索更多语系和语言，限制了发现的普适性。为解决上述问题，我们首先构建新基准Rephrased Multilingual LAMA（RML-LAMA），包含每个事实的高质量完形填空式多语言平行查询；然后提出Multilingual Integrated Gradients with Uncertainty Estimation（MATRICE）方法，量化知识定位中跨查询和语言的不确定性。大量实验表明，我们的方法能准确地定位语言无关知识神经元。我们还进一步探究了这些神经元在跨语言知识编辑、知识增强和新知识注入中的作用。
英文摘要：
Large language models (LLMs) have learned vast amounts of factual knowledge through self-supervised pre-training on large-scale corpora. Meanwhile, LLMs have also demonstrated excellent multilingual capabilities, which can express the learned knowledge in multiple languages. However, the knowledge storage mechanism in LLMs still remains mysterious. Some researchers attempt to demystify the factual knowledge in LLMs from the perspective of knowledge neurons, and subsequently discover language-agnostic knowledge neurons that store factual knowledge in a form that transcends language barriers. However, the preliminary finding suffers from two limitations: 1) High Uncertainty in Localization Results. Existing study only uses a prompt-based probe to localize knowledge neurons for each fact, while LLMs cannot provide consistent answers for semantically equivalent queries. Thus, it leads to inaccurate localization results with high uncertainty. 2) Lack of Analysis in More Languages. The study only analyzes language-agnostic knowledge neurons on English and Chinese data, without exploring more language families and languages. Naturally, it limits the generalizability of the findings. To address aforementioned problems, we first construct a new benchmark called Rephrased Multilingual LAMA (RML-LAMA), which contains high-quality cloze-style multilingual parallel queries for each fact. Then, we propose a novel method named Multilingual Integrated Gradients with Uncertainty Estimation (MATRICE), which quantifies the uncertainty across queries and languages during knowledge localization. Extensive experiments show that our method can accurately localize language-agnostic knowledge neurons. We also further investigate the role of language-agnostic knowledge neurons in cross-lingual knowledge editing, knowledge enhancement and new knowledge injection.
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。
