// the take
Awesome-LLM-RAG is a link-list README tracking academic papers, workshops, and books on retrieval-augmented generation for LLMs, grouped by sub-topic (instruction tuning, in-context learning, embeddings, evaluation, optimization, etc). It's built for researchers and ML engineers who want a reading list of RAG literature, not developers looking for a library to install.
The taxonomy is genuinely useful — splitting papers into retrieval-enhanced LLMs, instruction tuning, in-context learning, evaluation, and optimization makes it easier to locate relevant prior work than a flat chronological dump. It's actively maintained with 2026 papers and workshops still being added, so it isn't a stale 2023 snapshot like most awesome-RAG lists. Entries link straight to both the arXiv paper and the implementation's GitHub repo where one exists, saving a search step.
The 'Resources' and 'RAG Search' sections have been flooded with unrelated self-promotional product listings (desktop apps, MCP servers, memory frameworks) that have nothing to do with the academic papers the list is meant to track — classic awesome-list PR spam, and a sign there's no real gatekeeping on contributions anymore. It's a bare link list with one-line descriptions at best; there's no synthesis of which techniques actually work or how they compare, so you still have to read every paper yourself. There are no 'last verified' markers or pruning, so pre-2023 methods that have since been superseded sit next to 2026 work with no signal of which is still relevant. And there's nothing to install or run — it's pure documentation, so none of the usual repo-health signals (issue triage, releases, test coverage) exist to tell you whether the maintainer is still curating or just auto-merging PRs.
We read the README, the directory tree and a key source file in November 2026. Nobody pays to be listed. How we judge →