santosardr/riskernel

RIS-Kernel: A Model-Agnostic Architecture for Long-Context LLM Inference via Sparse Attention

3 best points · 2 comments · 2 Hacker News posts · 74 GitHub stars · View on GitHub

Hacker News discussions

• RIS (3 points, 2 comments, by r2ob, 2026-08-16)
• RIS-Kernel: Running 64k context LLMs on CPU via sparse attention (2 points, 0 comments, by santosardr, 2026-05-31)