RAG isn't just for chatbots anymore

What it does

This repository is the companion bibliography for the survey paper Retrieval-Augmented Generation for AI-Generated Content: A Survey. It collects research papers and sorts them into a taxonomy of RAG Foundations, Enhancements, and Applications. The scope stretches well beyond large language models into diffusion models, code generators, video captioners, and molecular design.

The interesting bit

Most RAG discourse stops at vector search for chatbots; this survey treats retrieval as a cross-modal building block. The taxonomy organizes papers by where retrieval plugs in—query, latent space, logits, or speculative decoding—rather than treating it as a black-box afterthought. You will find retrieval-augmented diffusion for images and motion, audio captioning via similar-caption retrieval, and even protein-ligand generation frameworks.

Key highlights

• Taxonomy covers Foundations (query-based, latent, logit, speculative), Enhancements (input, retriever, generator, result), and Applications.
• Multimodal coverage includes text-to-image (Re-Imagen, kNN-Diffusion), text-to-audio, video captioning, code completion, and 3D/molecular generation.
• Anchored to a living arXiv paper the authors commit to updating as the field grows.
• Roughly 100+ papers with direct arXiv, DOI, or OpenReview links.

Caveats

• The README is essentially a static paper list with no summaries, code, or interactive filtering.
• Several sections are truncated in the provided source, so the full depth is only partially visible.

Verdict

Bookmark this if you are building multimodal generative systems or writing a literature review. Skip it if you need runnable frameworks or distilled implementation notes.