Magic Mushroom: A Customizable Benchmark for Fine-grained Analysis of Retrieval Noise Erosion in RAG Systems
Main:9 Pages
32 Figures
Bibliography:5 Pages
9 Tables
Appendix:16 Pages
Abstract
Retrieval-Augmented Generation (RAG) systems enhance Large Language Models (LLMs) by incorporating external retrieved information, mitigating issues such as hallucination and outdated knowledge.
View on arXiv@article{zhang2025_2506.03901,
title={ Magic Mushroom: A Customizable Benchmark for Fine-grained Analysis of Retrieval Noise Erosion in RAG Systems },
author={ Yuxin Zhang and Yan Wang and Yongrui Chen and Shenyu Zhang and Xinbang Dai and Sheng Bi and Guilin Qi },
journal={arXiv preprint arXiv:2506.03901},
year={ 2025 }
}Comments on this paper