About the acceleration frame work.

[Yxxxb]

Hi!

I wonder if you have tried to deploy any acceleration or memory saving framework for InternLM-XComposer? like vLLM. If so can you provide code or possible solution?

Thanks!