liunian-Jay/Awesome-RAG

What it solves

Keeping up with the fast-moving field of Retrieval-Augmented Generation (RAG) is hard. This repository gathers recent academic papers, surveys, evaluation datasets, and related open-source projects into one place, so researchers and practitioners can quickly find what’s new in RAG without searching multiple sources.

How it works

The repo is organized as a structured list. It groups papers by year and month (2024, 2025, 2026), with links to arXiv and sometimes code. It also lists accepted papers at major conferences (ACL, EMNLP, ICML, ICLR, NIPS) and provides a separate section for evaluation datasets like HotpotQA and TriviaQA. A “Latest News” section highlights the maintainer’s own contributions and updates. Survey papers and related projects (e.g., LightRAG, RAGFlow) are collected under a Resources section.

Who it’s for

Academics, students, and industry practitioners who want to track RAG research trends, find relevant papers for their work, or discover benchmarks and tools. The README explicitly welcomes contributors and invites researchers to submit pull requests to update paper information.

Highlights

• Broad chronological coverage: papers from 2024 through 2026, organized by month for easy browsing.
• Includes both top-conference accepted papers and preprints, with direct links to PDFs and code repositories.
• Curated list of standard RAG evaluation datasets (HotpotQA, SQuAD, TriviaQA, etc.).
• A dedicated section for survey papers summarizing the field’s landscape.
• Maintainers actively share their own research (e.g., GainRAG, CoCoA, QAgent) and small tools (MU-GOT, AgenticRAG-RL) as part of the updates.