Xnhyacinth/Awesome-LLM-Long-Context-Modeling

Organizes research across 16+ technical categories including efficient attention mechanisms (sparse, linear, hierarchical), state space models, retrieval-augmented generation, and compression techniques for long-context processing. Maintains curated listings of papers on length extrapolation, KV cache optimization, long-horizon video understanding, and chain-of-thought reasoning, with weekly updates tracking emerging work. Includes links to a formal comprehensive survey paper and companion research repository for deeper exploration of architectural innovations and benchmark evaluations.

1,931 stars. Actively maintained with 16 commits in the last 30 days.