vllm
vllm copied to clipboard
Can you choose which GPU to use. like tf inference device_map="cuda:0"
As the title suggests