Evaluating Sparse and Dense Retrieval in Retrieval-Augmented Generation Systems: A Study
Abstract
With the rapid advancement of large-scale pre-trained models, their capabilities have found widespread applications across various fields, leading to increasing demand from individual users.However, personal users often face challenges such as concerns over data security and limitations in hardware resources, making the development and deployment of these models a complex task.Retrieval-Augmented Generation (RAG) systems offer a promising solution by enhancing a model's ability to acquire knowledge through the integration of external knowledge bases, without the need for extensive training.Nevertheless, RAG systems employ a variety of retrieval algorithms, and different configurations can significantly impact both resource consumption and performance.As a result, choosing the right retrieval algorithm has become a critical area of research.This study evaluates and compares sparse and dense retrieval algorithms, aiming to identify how RAG system performance can be optimized under varying resource constraints and user requirements.The experimental results demonstrate that a well-matched combination of Large-scale Language Model and retrieval algorithm can maximize system performance under specific hardware conditions, offering new strategies for individual users to efficiently leverage large models.
Read the paper
Related studies
- Knowledge Management Perspective of Generative Artificial Intelligence
- Navigating Generative Artificial Intelligence Promises and Perils for Knowledge and Creative Work
- Characterizing the evolving research on enterprise content management
- Large language models and multimodal foundation models for precision oncology
- From Generative AI to Extended Reality: Multidisciplinary Perspectives on the Challenges, Opportunities, and Future of Educational Computing
- Generative Artificial Intelligence: A New Engine for Advancing Environmental Science and Engineering
Go deeper in Evidence Hub
Evidence Hub adds a study data sheet (design, sample size, evidence level, key numbers), an AI summary and a chat grounded in the paper's text.