This repository provides a comprehensive, structured guide to learning Retrieval-Augmented Generation (RAG) from foundational concepts to advanced techniques. It targets developers and researchers seeking to build and evaluate LLM applications that leverage external knowledge, offering a curated roadmap and practical toolkits.

How It Works

The guide breaks down RAG into core components: indexing, retrieval, augmentation, and generation. It explains the necessity of RAG for addressing LLM limitations like knowledge cutoffs and hallucinations. The content is organized into distinct courses and toolkits, covering RAG basics, evaluation metrics, and various implementation frameworks.

Quick Start & Requirements

This repository is a guide, not a runnable application. It links to external resources and libraries for implementation. Key frameworks mentioned include LangChain, Llama Index, Haystack, and CrewAI. Vector databases like Chroma, Qdrant, and Milvus are also featured.

Highlighted Details

• Curated list of RAG frameworks, vector databases, data extraction tools, chunking libraries, rerankers, and agentic RAG orchestrators.
• Detailed sections on RAG evaluation, including metrics and implementations with RAGAS and DeepEval.
• Extensive collection of survey papers on RAG, covering general, graph RAG, agentic RAG, and evaluation aspects.
• Practical examples for implementing RAG with LangChain, CrewAI, and for specific use cases like website or YouTube video RAG.

Maintenance & Community

The repository is maintained by KalyanKS-NLP. Links to external resources and community channels are provided within the guide's sections.

Licensing & Compatibility

The repository itself does not appear to have a specific license mentioned in the README. The linked libraries and frameworks will have their own individual licenses, which users must consult for compatibility and usage restrictions.

Limitations & Caveats

As a guide, this repository does not provide executable code for a complete RAG system. Users will need to integrate the various tools and concepts presented to build their own solutions. The effectiveness of specific implementations will depend on the chosen libraries and datasets.