Using a remote hosted vLLM

[anupkashyap]

I'm trying to use LiteLLM with a vllm hosted on a GPU instance. From the docs, I see examples of vLLM running on the same instance as the literLLM installation and that it requires vllm dependency (which requires cuda). However since I am just making a call to a remote cluster that hosts the vLLM, I doubt that there is any GPU requirement on the liteLLM installation. I was able to get around this issue by modifying the "custom" provider to make calls to the server and process the response format. Please help me if there is a better way to do this.