rag-in-action
End-to-end RAG system design, evaluation, and optimization. 极客时间RAG训练营，RAG 10大组件全面拆解，4个实操项目吃透 RAG 全流程。RAG的落地，往往是面向业务做RAG
What is rag-in-action?
End-to-end RAG system design, evaluation, and optimization. 极客时间RAG训练营，RAG 10大组件全面拆解，4个实操项目吃透 RAG 全流程。RAG的落地，往往是面向业务做RAG，而不是反过来面向RAG做业务。这就是为什么我们需要针对不同场景、不同问题做针对性的调整、优化和定制化。魔鬼全在细节中，我们深入进去探究。
Overview of rag-in-action
What it is
This repository is a code repository for a practical course on RAG (Retrieval-Augmented Generation) system development based on DeepSeek. It implements a complete retrieval-augmented generation system. The project is modular, covering 10 components of RAG: data loading, document chunking, embedding, vector database, pre-retrieval, indexing, post-retrieval, generation, evaluation, and advanced RAG techniques. The course is offered by 极客时间 (Geek Time) and a companion book 'RAG实战课' has been published.
Key features
- Modular design with 10 independent modules covering the full RAG pipeline - Support for both LangChain and LlamaIndex frameworks - Support for GPU (NVIDIA, CUDA) and CPU environments - Includes simple RAG, advanced RAG (Graph RAG, Multi-Agent), and evaluation using RAGAS and TruLens - Provides environment configuration files for Ubuntu, MacOS, and Windows
Use cases
- Learning and practicing RAG system design and optimization - Building production-ready RAG applications for business scenarios - Experimenting with different retrieval, indexing, and generation strategies - Evaluating and improving RAG system performance
Getting started
bash # Create virtual environment (Ubuntu GPU + LangChain) python -m venv venv-rag-langchain source venv-rag-langchain/bin/activate # Or using conda conda create -n venv-rag-langchain python=3.10.12 conda activate venv-rag-langchain # Install dependencies pip install -r 91-环境-Environment/requirements_langchain_20250413_Ubuntu-with-GPU.txt
License and pricing
The project is open source under the MIT License. The course is offered by 极客时间 (Geek Time) with a separate pricing model (not specified in the repository).
Based on the project's public README and repository metadata.
