On The Reproducibility Limitations of RAG Systems

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wang, Baiqiang, Zhao, Dongfang, Tallent, Nathan R, Guo, Luanzheng
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911171872817152
author Wang, Baiqiang
Zhao, Dongfang
Tallent, Nathan R
Guo, Luanzheng
author_facet Wang, Baiqiang
Zhao, Dongfang
Tallent, Nathan R
Guo, Luanzheng
contents Retrieval-Augmented Generation (RAG) is increasingly employed in generative AI-driven scientific workflows to integrate rapidly evolving scientific knowledge bases, yet its reliability is frequently compromised by non-determinism in their retrieval components. This paper introduces ReproRAG, a comprehensive benchmarking framework designed to systematically measure and quantify the reproducibility of vector-based retrieval systems. ReproRAG investigates sources of uncertainty across the entire pipeline, including different embedding models, precision, retrieval algorithms, hardware configurations, and distributed execution environments. Utilizing a suite of metrics, such as Exact Match Rate, Jaccard Similarity, and Kendall's Tau, the proposed framework effectively characterizes the trade-offs between reproducibility and performance. Our large-scale empirical study reveals critical insights; for instance, we observe that different embedding models have remarkable impact on RAG reproducibility. The open-sourced ReproRAG framework provides researchers and engineers productive tools to validate deployments, benchmark reproducibility, and make informed design decisions, thereby fostering more trustworthy AI for science.
format Preprint
id arxiv_https___arxiv_org_abs_2509_18869
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle On The Reproducibility Limitations of RAG Systems
Wang, Baiqiang
Zhao, Dongfang
Tallent, Nathan R
Guo, Luanzheng
Distributed, Parallel, and Cluster Computing
Retrieval-Augmented Generation (RAG) is increasingly employed in generative AI-driven scientific workflows to integrate rapidly evolving scientific knowledge bases, yet its reliability is frequently compromised by non-determinism in their retrieval components. This paper introduces ReproRAG, a comprehensive benchmarking framework designed to systematically measure and quantify the reproducibility of vector-based retrieval systems. ReproRAG investigates sources of uncertainty across the entire pipeline, including different embedding models, precision, retrieval algorithms, hardware configurations, and distributed execution environments. Utilizing a suite of metrics, such as Exact Match Rate, Jaccard Similarity, and Kendall's Tau, the proposed framework effectively characterizes the trade-offs between reproducibility and performance. Our large-scale empirical study reveals critical insights; for instance, we observe that different embedding models have remarkable impact on RAG reproducibility. The open-sourced ReproRAG framework provides researchers and engineers productive tools to validate deployments, benchmark reproducibility, and make informed design decisions, thereby fostering more trustworthy AI for science.
title On The Reproducibility Limitations of RAG Systems
topic Distributed, Parallel, and Cluster Computing
url https://arxiv.org/abs/2509.18869