https://github.com/arian-ott/rag-benchmark

https://github.com/arian-ott/rag-benchmark

Last synced: 27 days ago
JSON representation

• Host: GitHub
• URL: https://github.com/arian-ott/rag-benchmark
• Owner: Arian-Ott
• License: agpl-3.0
• Created: 2024-08-16T12:40:57.000Z (3 months ago)
• Default Branch: main
• Last Pushed: 2024-09-08T20:48:46.000Z (about 2 months ago)
• Last Synced: 2024-09-08T21:57:38.460Z (about 2 months ago)
• Language: Python
• Size: 4.2 MB
• Stars: 1
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • Codeowners: CODEOWNERS

Awesome Lists containing this project

README

        
# RAG Benchmark

This repository, titled **RAG Benchmark**, contains all programming 

examples related to the first practical exam paper of the DHBW Heilbronn

(Baden-Wuerttemberg 

Cooperative 

State University). The content of this repository was designed all by myself. 

The focus of this repository is on benchmarking different approaches within the Retrieval-Augmented Generation (RAG) framework.

## Contents

The repository is structured to provide a comprehensive benchmarking platform for three distinct RAG variants:

1. **Naive RAG**:

   - A basic implementation of RAG, providing a foundational benchmark for comparison.

   

2. **Advanced RAG**:

   - An enhanced version of RAG, incorporating optimisations and improvements over the Naive RAG. This version is expected to offer better performance and more accurate results.

3. **Modular RAG**:

   - A modular approach to RAG, where various components are independently developed and benchmarked. This version is designed for flexibility, allowing for easy substitution and testing of individual modules.

## Purpose

The primary aim of this repository is to offer a detailed comparison of these three RAG implementations. By benchmarking these versions against each other, the goal is to identify strengths, weaknesses, and potential areas for further optimisation in practical applications.

## Usage

Each RAG variant is contained within its respective directory, with clear instructions on how to execute the benchmarking tests. The repository is designed to be accessible to students and practitioners, providing both practical examples and a solid foundation for further development and experimentation.

## Getting started

---

This project is part of the Practical Work 1 requirements at the DHBW and serves as a hands-on educational tool for understanding and improving RAG-based systems.