This repository serves as a comprehensive catalog of research papers on Retrieval-Augmented Generation (RAG), organized according to a taxonomy presented in the paper "Retrieval-Augmented Generation for AI-Generated Content: A Survey." It is intended for researchers and practitioners in the field of AI-generated content who need a structured overview of RAG techniques, enhancements, and applications.

How It Works

The repository categorizes RAG papers into foundational methods (e.g., query-based, latent representation-based, logit-based, speculative), enhancements (e.g., input, retriever, generator, result, pipeline), and applications across various domains like text, code, audio, image, video, 3D, knowledge, and science. This structured approach allows users to navigate the rapidly evolving RAG landscape and identify relevant research.

Quick Start & Requirements

This repository is a curated list of academic papers and does not require installation or execution. It provides links to arXiv, official websites, and other relevant resources for each paper.

Highlighted Details

• Extensive taxonomy covering RAG foundations, enhancements, and applications.
• Links to over 300 research papers with DOIs or arXiv IDs.
• Includes a "Benchmark" section with evaluations for RAG systems.
• Provides citation details for the survey paper.

Maintenance & Community

The repository is associated with the survey paper "Retrieval-Augmented Generation for AI-Generated Content: A Survey" by Zhao et al. (2024). Updates to the paper and repository are ongoing to reflect the field's rapid growth.

Licensing & Compatibility

The repository itself is likely under a permissive license (e.g., MIT, Apache 2.0, as is common for GitHub repos), but the content consists of links to research papers, each with its own copyright and distribution terms. Compatibility for commercial use depends on the licenses of the linked papers.

Limitations & Caveats

This repository is a curated list and does not provide code implementations or direct access to the full text of all papers. Users must follow the provided links to access the original research.