Efficient Multi-Lora Deployment via Shared KV-Cache with Task-Adaptive Tokens
Jie Ou,Jinyu Guo,Shuaihong Jiang,Wenhong Tian