Spaces:

huggingchat

/

chat-ui

like 1.02k

Paused

The model is not running in the Space itself, the Space is just a webapp that proxies calls to the HF inference API

It's running in GPUs

any details on the inference setup?

g5 instances from aws currently

running https://github.com/huggingface/text-generation-inference

@julien-c cool glad to you hear you don't need A100s to get speed like that. Using the base model https://huggingface.co/OpenAssistant/oasst-sft-6-llama-30b-xor and not special quantization/distillation on top??

g5 instances from aws currently

Hello @julien-c ,

could you please kindly share the parameters used for starting the server. I also happened to use g5 instance for inference, speed good, but not as good as this demo.

Do you run https://github.com/huggingface/text-generation-inference ?

yes, not docker but cli