arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2508.01302cs.CLcs.CE
收藏

对齐语言模型与实时知识编辑

Aligning Language Models with Real-time Knowledge Editing

National Key Laboratory for Multimedia Information Processing, Peking University School of Computer Science, Peking University(国家多媒体信息处理重点实验室,北京大学计算机学院,北京大学)

Chenming Tang, Yutong Yang, Kexue Wang, Yunfang Wu

AI总结:

本文提出CRAFT动态知识编辑数据集及KEDAS方法,通过多样编辑增强和自适应推理提升知识编辑性能,推动从静态更新向动态演化转变。

摘要:

知识编辑旨在高效修改语言模型中的过时知识并保留原有能力。主流知识编辑数据集多为静态,难以跟上现实世界知识的演变。本文引入CRAFT,一个持续更新的现实世界知识编辑数据集,评估模型在时间局部性、常识局部性、复合可移植性和别名可移植性等方面的表现,为知识编辑提供全面挑战性的评估。针对灵活的实时知识编辑,我们提出KEDAS,一种具有多样编辑增强和自适应后对齐推理的新知识编辑对齐范式,在CRAFT和传统数据集上相比先前方法表现出显著性能提升。我们希望这项工作能成为推动知识编辑从静态更新向动态演化转变的催化剂。

英文摘要:

Knowledge editing aims to modify outdated knowledge in language models efficiently while retaining their original capabilities. Mainstream datasets for knowledge editing are predominantly static and fail to keep in pace with the evolving real-world knowledge. In this work, we introduce CRAFT, an ever-evolving real-world dataset for knowledge editing. It evaluates models on temporal locality, common-sense locality, composite portability and alias portability, providing a comprehensive and challenging evaluation for knowledge editing, on which previous methods hardly achieve balanced performance. Towards flexible real-time knowledge editing, we propose KEDAS, a novel paradigm of knowledge editing alignment featuring diverse edit augmentation and self-adaptive post-alignment inference, exhibiting significant performance gain on both CRAFT and traditional datasets compared to previous methods. We hope this work may serve as a catalyst for shifting the focus of knowledge editing from static update to dynamic evolution.

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。

↑