Sim-LLM: Optimizing LLM Inference at the Edge through Inter-Task KV Reuse

Luo, Ruikun, Gu, Changwei, He, Qiang · Advances in Neural Information Processing Systems 38 · 2025