提出非破坏性协作知识编辑框架，解决多方协同改写模型时的冲突与遗忘问题。
CollabEdit: Towards Non-destructive Collaborative Knowledge Editing
- 设计新型模型融合机制，实现隐私保护下的持续知识聚合。
- 在两个数据集上验证，相比破坏性方法性能下降减少超过40%。
- 适合需要安全协作更新大模型知识的科研与企业场景。
多方协作学习大语言模型（LLMs）已成为利用各方私有数据提升效率与保障隐私的新范式。与此同时，知识编辑（KE）因能显式操控模型行为而备受关注，但多方协作下的知识编辑（即多方知识修改在隐私保护与持续性条件下聚合）尚未被研究。本文首次系统探究协作知识编辑，识别出三大挑战：知识重叠、知识冲突与知识遗忘。为此，提出非破坏性协作知识编辑框架COLLABEDIT，采用创新模型融合机制，模拟全局知识编辑行为，有效避免性能显著下降。在两个标准数据集上的大量实验表明，相比其他破坏性基线方法，COLLABEDIT表现更优，且为应对三大挑战提供了可行方案并展望未来应用。代码已开源：https://github.com/LINs-lab/CollabEdit。
原文摘要 · Abstract (English)
Collaborative learning of large language models (LLMs) has emerged as a new paradigm for utilizing private data from different parties to guarantee efficiency and privacy. Meanwhile, Knowledge Editing (KE) for LLMs has also garnered increased attention due to its ability to manipulate the behaviors of LLMs explicitly, yet leaves the collaborative KE case (in which knowledge edits of multiple parties are aggregated in a privacy-preserving and continual manner) unexamined. To this end, this manuscript dives into the first investigation of collaborative KE, in which we start by carefully identifying the unique three challenges therein, including knowledge overlap, knowledge conflict, and knowledge forgetting. We then propose a non-destructive collaborative KE framework, COLLABEDIT, which employs a novel model merging mechanism to mimic the global KE behavior while preventing the severe performance drop. Extensive experiments on two canonical datasets demonstrate the superiority of COLLABEDIT compared to other destructive baselines, and results shed light on addressing three collaborative KE challenges and future applications. Our code is available at https://github.com/LINs-lab/CollabEdit.
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。