Running 1 RIS-Kernel Long Context Inference Demo 🔬 Run long‑context queries on PDFs with O(N log N) sparse attention