Welcome to the RAG-wiki!

Work attribution

This wiki is part of the work completed during an internship at IRT Saint Exupéry, France (Summer 2024). The pages here have been contributed by the intern and duplicated to this GitHub repository, with content dated up to August 2024.

The wiki serves as a curated collection of information, including recommendations for blogs, images, and papers worth exploring.

Please note that any new pages added to this repository after August 2024 are independent of the IRT Saint Exupéry internship.

Overview

The RAG-wiki offers an in-depth look at Retrieval Augmented Generation (RAG), focusing on its core concepts and practical applications. By integrating retrieval systems, RAG significantly enhances language model performance.

Explore the stages of a RAG pipeline, from data collection to evaluation, with clear, step-by-step guidance and best practices. Special attention is given to Multimodal RAG, which leverages various data types like text, images, and tables.

Looking ahead, the wiki will feature surveys on advanced RAG techniques and provide insights into industrial use cases, practical implementations, and emerging best practices. Dive in to stay ahead in the evolving field of RAG.

If you find this resource valuable, consider giving the repository a star to support its ongoing development!

Table of contents

• Home

• Introducing RAG

• Stages of RAG

  • S01_Data Collection

  • S02_Data Parsing

    • PDF Parsing
    • PDF Parsing Benchmark

  • S03_Data Cleaning

  • S04_Chunking

    • Chunking strategies
    • How to select chunking strategies

  • S05_Embedding

    • How to choose a text embedding model

  • S06_Indexing

    • How to choose a vector database

  • S07_Storing

  • S08_Retrieval

  • S09_Augmentation

  • S10_Generation

  • S11_Evaluation

• Advanced RAG

• Multimodal RAG

  • Multimodal RAG approaches

• Unimodal and Multimodal RAG pipelines examples

Ongoing pages

• Different WebUI frameworks for RAG
• Detailed illustrations of each technique of advanced RAG
• RAG best practices
• chunking in real use case
• Agentic RAG
• Modular RAG
• Fine-tuning of different RAG components