Fetching the paper…

InstInfer: In-Storage Attention Offloading for Cost-Effective Long-Context LLM Inference · Around