can vllm use autogptq compressed model like llama-2-70b-chat-gptq

[fancyerii]

I want to use llama-2-70b-chat-gptq. Can it be used in vllm?

[Tostino]

No, you can use awq quantized models though. GPTQ is not supported at this point in time.

[aarnphm]

There is a running PR #916

[fancyerii]

@aarnphm is this pr ready to use?