Fetching the paper…

AlpaServe: Statistical Multiplexing with Model Parallelism for Deep Learning Serving · Around