提出统一框架，系统研究大模型知识更新机制
KnowledgeSmith: Uncovering Knowledge Updating in LLMs with Model Editing and Unlearning
- 将编辑与遗忘建模为约束优化问题，统一分析框架
- 发现模型更新行为与人类不同，存在一致性与容量权衡
- 适合关注模型可信赖性与可扩展性的研究人员
知识编辑与机器遗忘是大语言模型保持更新的两种主流方法，但其知识更新机制仍缺乏系统探索，主要受限于数据不足、孤立且规模有限的评估。例如，大模型在知识修改上是否类似人类？随着训练数据增加，编辑与遗忘有何差异？本文提出KnowledgeSmith，一个统一框架，用于系统理解大模型的知识更新机制。我们首先将编辑与遗忘形式化为同一类约束优化问题；随后设计自动数据集生成器，在多图层级与多数据尺度下提供结构化干预，实现对不同修改策略如何传播至模型知识的受控研究。大量实验揭示了知识传播、可塑性缩放、一致性与鲁棒性方面的细致洞察。例如，结果表明大模型在不同知识层级上的更新行为不似人类，且存在一致性-容量权衡。希望本研究能为更可靠、可扩展的更新策略设计提供启示。代码：https://github.com/AIFrontierLab/KnowledgeSmith.git
原文摘要 · Abstract (English)
Knowledge editing and machine unlearning are two popular approaches for large language models (LLMs) to stay up-to-date. However, the knowledge updating mechanism of LLMs remains largely unexplored due to insufficient, isolated, and small-scale evaluation. For instance, are LLMs similar to humans in modifying certain knowledge? What differs editing and unlearning as training data increases? This paper proposes KnowledgeSmith, a unified framework to systematically understand the updating mechanism of LLMs. We first cast editing and unlearning as instances of one constrained optimization problem. Then, we propose an automatic dataset generator that provides structured interventions across multiple graph levels and data scales, enabling controlled studies of how different modification strategies propagate through model knowledge. Extensive experiments demonstrate nuanced insights over knowledge propagation, plasticity scaling, consistency, and robustness. For instance, our results show that LLMs do not exhibit similar updating as humans for different levels of knowledge, and there exists consistency-capacity trade-off. We hope our findings can offer suggestions to the design of more reliable and scalable strategies. Code: https://github.com/AIFrontierLab/KnowledgeSmith.git
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。