Saved in:
Bibliographic Details
Main Authors: Orbach, Matan, Eytan, Ohad, Sznajder, Benjamin, Gera, Ariel, Boni, Odellia, Kantor, Yoav, Bloch, Gal, Levy, Omri, Abraham, Hadas, Barzilay, Nitzan, Shnarch, Eyal, Factor, Michael E., Ofek-Koifman, Shila, Ta-Shma, Paula, Toledo, Assaf
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2505.03452
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911345820041216
author Orbach, Matan
Eytan, Ohad
Sznajder, Benjamin
Gera, Ariel
Boni, Odellia
Kantor, Yoav
Bloch, Gal
Levy, Omri
Abraham, Hadas
Barzilay, Nitzan
Shnarch, Eyal
Factor, Michael E.
Ofek-Koifman, Shila
Ta-Shma, Paula
Toledo, Assaf
author_facet Orbach, Matan
Eytan, Ohad
Sznajder, Benjamin
Gera, Ariel
Boni, Odellia
Kantor, Yoav
Bloch, Gal
Levy, Omri
Abraham, Hadas
Barzilay, Nitzan
Shnarch, Eyal
Factor, Michael E.
Ofek-Koifman, Shila
Ta-Shma, Paula
Toledo, Assaf
contents Optimizing Retrieval-Augmented Generation (RAG) configurations for specific tasks is a complex and resource-intensive challenge. Motivated by this challenge, frameworks for RAG hyper-parameter optimization (HPO) have recently emerged, yet their effectiveness has not been rigorously benchmarked. To fill this gap, we present a comprehensive study involving five HPO algorithms over five datasets from diverse domains, including a newly curated real-world product documentation dataset. Our study explores the largest RAG HPO search space to date that includes full grid-search evaluations, and uses three evaluation metrics as optimization targets. Analysis of the results shows that RAG HPO can be done efficiently, either greedily or with random search, and that it significantly boosts RAG performance for all datasets. For greedy HPO approaches, we show that optimizing model selection first is preferable to the common practice of following the RAG pipeline order during optimization.
format Preprint
id arxiv_https___arxiv_org_abs_2505_03452
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle An Analysis of Hyper-Parameter Optimization Methods for Retrieval Augmented Generation
Orbach, Matan
Eytan, Ohad
Sznajder, Benjamin
Gera, Ariel
Boni, Odellia
Kantor, Yoav
Bloch, Gal
Levy, Omri
Abraham, Hadas
Barzilay, Nitzan
Shnarch, Eyal
Factor, Michael E.
Ofek-Koifman, Shila
Ta-Shma, Paula
Toledo, Assaf
Computation and Language
Artificial Intelligence
Machine Learning
Optimizing Retrieval-Augmented Generation (RAG) configurations for specific tasks is a complex and resource-intensive challenge. Motivated by this challenge, frameworks for RAG hyper-parameter optimization (HPO) have recently emerged, yet their effectiveness has not been rigorously benchmarked. To fill this gap, we present a comprehensive study involving five HPO algorithms over five datasets from diverse domains, including a newly curated real-world product documentation dataset. Our study explores the largest RAG HPO search space to date that includes full grid-search evaluations, and uses three evaluation metrics as optimization targets. Analysis of the results shows that RAG HPO can be done efficiently, either greedily or with random search, and that it significantly boosts RAG performance for all datasets. For greedy HPO approaches, we show that optimizing model selection first is preferable to the common practice of following the RAG pipeline order during optimization.
title An Analysis of Hyper-Parameter Optimization Methods for Retrieval Augmented Generation
topic Computation and Language
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2505.03452