Request-Adaptive Execution for Efficient Hybrid LLM Inferencewith Fahao Chen, Qi Zhang, Peng Li, Dongxiao Yu +11 versionW_adhtfndxv1 · current