解决大模型知识编辑时的关联信息丢失问题，让修改更精准。
Causal Path Alignment: Anchoring the Optimization Trajectory for Controllable In-Parameter Knowledge Editing
- 通过因果路径对齐，强制参数更新经过关系感知的中间状态。
- 在多个模型上显著提升关系准确性，且副作用极小。
- 可直接接入现有编辑工具，适合需要可靠知识更新的研究者。
知识编辑对高效更新大语言模型（LLM）的参数记忆至关重要，使模型能在动态环境中持续演化。然而，主流的参数内知识编辑方法存在主体主导记忆干扰：修改特定事实会无意中破坏同一主体相关的整体结构知识。我们诊断其根源为捷径学习病理，即优化目标过度拟合主体表示，绕过了必要的关系上下文。为此，我们提出因果路径对齐（CPA），一个旨在将优化轨迹锚定于有效因果路径的原理性框架。CPA 强制参数更新通过关系感知的中间状态，防止上下文依赖关系被消除。在多种 LLM 架构上的实验表明，CPA 能一致地消除捷径，显著提升关系特异性，同时表现出极小的副作用。此外，CPA 可作为模型无关的插件集成至现有编辑器，为可靠、可信的参数内知识编辑铺平道路。
原文摘要 · Abstract (English)
Knowledge editing is pivotal for efficiently updating the parametric memory of Large Language Models (LLMs), enabling them to function as evolving agents in dynamic environments. However, mainstream in-parameter knowledge editing approaches suffer from Subject-Dominant Memory Interference: modifying a specific fact inadvertently corrupts the broader structural knowledge associated with the same subject within LLMs. We diagnose the root cause as a shortcut learning pathology, where the optimization objective overfits subject representations while bypassing the essential relational context. To rectify this, we propose Causal Path Alignment (CPA), a principled framework designed to anchor the optimization trajectory to valid causal pathways. CPA enforces parameter updates to route through relation-aware intermediate states, thereby preventing the erasure of contextual dependencies. Experimental results across diverse LLM backbones demonstrate that CPA consistently eliminates the shortcut, significantly improving relation specificity while exhibiting minimal side-effects. Moreover, CPA serves as a model-agnostic plug-in for existing editors, paving the way for reliable and trustworthy in-parameter knowledge editing.
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。