TailorKV: A Hybrid Framework for Long-Context Inference via Tailored KV Cache Optimization

Yao, Dingyu, Shen, Bowen, Lin, Zheng · arXiv · 2025