---
id: 20260930-T0-11
title: "扩散模型里的RAG冲突：用轨迹方差可视化并检测知识打架"
title_en: "Visualizing and Detecting RAG Conflicts in Diffusion Models via Trajectory Variance"
url: https://ai.daily.yangsir.net/daily/20260930-T0-11
issue_date: 2026-09-30
publish_date: 2026-09-29T04:00:00.000Z
category: research
source_name: "arXiv cs.CL (NLP)"
source_url: https://arxiv.org/abs/2609.31684
---

# 扩散模型里的RAG冲突：用轨迹方差可视化并检测知识打架

检索增强生成（RAG）在离散扩散语言模型中会引发一种特定失效模式：当检索到的上下文与模型参数知识相矛盾时，迭代去噪过程会变成两种知识之间的战场。研究通过轨迹方差来可视化和检测这种 RAG 冲突，让模型内部“听谁的”这一过程变得可观测。该方法为识别扩散语言模型中的知识冲突提供了检测手段，对构建 RAG 系统的开发者有直接价值：可以在生成过程中发现检索内容与模型固有知识不一致的情况，从而调整检索策略或提示设计，减少错误输出。

## English Version

**Visualizing and Detecting RAG Conflicts in Diffusion Models via Trajectory Variance**

Retrieval-Augmented Generation (RAG) introduces a specific failure mode in discrete diffusion language models: when retrieved context contradicts parametric knowledge, the iterative denoising process becomes a visible battleground between competing knowledge sources. This research uses trajectory variance to visualize and detect such RAG conflicts, making the model's internal 'which source wins' process observable. The method offers a detection tool for knowledge conflicts in diffusion language models, with direct value for developers building RAG systems: they can spot mismatches between retrieved content and the model's built-in knowledge during generation, then adjust retrieval strategies or prompt design to reduce erroneous outputs.

---

**来源**：[arXiv cs.CL (NLP)](https://arxiv.org/abs/2609.31684)

**详情页**：https://ai.daily.yangsir.net/daily/20260930-T0-11

---

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