Accelerating Prefilling for Long-Context LLMs via Sparse Pattern Sharing

Re-assign community