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