---
id: 20260923-T0-05
title: "长上下文模型会“中毒”：无关上下文越多，越找不到关键证据"
title_en: "Context Poisoning: Long-Context LLMs Lose Key Evidence as Irrelevant Text Grows"
url: https://ai.daily.yangsir.net/daily/20260923-T0-05
issue_date: 2026-09-23
publish_date: 2026-09-22T04:00:00.000Z
category: research
source_name: "arXiv cs.CL (NLP)"
source_url: https://arxiv.org/abs/2609.22101
---

# 长上下文模型会“中毒”：无关上下文越多，越找不到关键证据

一篇新论文提出“上下文中毒”现象：大模型虽然能处理越来越长的提示，但当无关或易混淆的内容增加时，定位并使用关键证据的能力会下降。作者将其建模为极值注意力干扰问题，从注意力机制层面解释这一退化。该研究对依赖长文档问答、法律与医疗检索等场景有直接参考价值，提示开发者不能只看上下文窗口大小，还要关注模型在噪声环境下的实际检索能力。

## English Version

**Context Poisoning: Long-Context LLMs Lose Key Evidence as Irrelevant Text Grows**

A new paper formalizes "context poisoning": although large language models can process increasingly long prompts, their ability to locate and use decisive evidence degrades as irrelevant or confusable context is added. The authors model this as extreme-value attention interference, explaining the degradation at the attention level. The work is directly relevant to long-document QA, legal and medical retrieval, and suggests developers should look beyond context window size to how models actually retrieve evidence under noise.

---

**来源**：[arXiv cs.CL (NLP)](https://arxiv.org/abs/2609.22101)

**详情页**：https://ai.daily.yangsir.net/daily/20260923-T0-05

---

*智语观潮 · Daily — https://ai.daily.yangsir.net/llms.txt*