提出新方法解决大模型编辑后依赖上下文的问题。
Uncovering Context Reliance in Unstructured Knowledge Editing
- 设计不依赖上下文的编辑框架，让模型聚焦局部知识
- 实验显示上下文缺失时召回率提升45.2%
- 适合需要稳定知识更新的部署场景
大型语言模型（LLMs）在现实世界中进行非结构化知识编辑至关重要，用于纠正和更新其内部参数化知识。本文重新审视以下一个词预测（NTP）为基础的编辑范式，发现其存在关键缺陷——上下文依赖性（Context Reliance），即编辑后知识过度依赖前序上下文，导致推理时若上下文缺失则无法召回。实验证明，在推理时添加原始上下文可恢复知识召回。理论分析表明，基于梯度的优化会自然地将知识绑定到特定上下文表征上。为此，我们提出简单有效的无上下文依赖编辑框架（COIN），鼓励模型关注局部知识而非记忆上下文模式。评估结果表明，COIN将上下文依赖性降低45.2%，编辑成功率比强基线高出23.6%，凸显缓解上下文依赖对实现稳健编辑的关键作用。
原文摘要 · Abstract (English)
Editing Large language models (LLMs) with real-world, unstructured knowledge is essential for correcting and updating their internal parametric knowledge. In this work, we revisit the fundamental next-token prediction (NTP) as a candidate paradigm for unstructured editing. We identify Context Reliance as a critical failure mode of NTP-based approaches, where knowledge acquired from edited text becomes highly dependent on its preceding context, leading to recall failures when that context is absent during inference. This hypothesis is supported by our empirical validation that prepending context during inference recovers knowledge recall. We further theoretically demonstrate that Context Reliance is an inherent consequence of gradient-based optimization, which tends to bind acquired knowledge to a specific aggregated contextual representation. To address this, we propose a simple yet effective COntext-INdependent editing framework (COIN), encouraging model to focus on knowledge within local scope rather than memorizing contextual patterns. Evaluations show that COIN reduces Context Reliance by 45.2% and outperforms strong baselines by 23.6% in editing success rate, highlighting the vital role of mitigating Context Reliance for robust 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。