How should vllm start it?
#24
by
SongXiaoMao
- opened
How should vllm start it?
they have not given support for mistral
https://huggingface.co/FlorianJc/Mistral-Nemo-Instruct-2407-vllm-fp8
If you want to use it now, you can install vllm from git or apply the patch manually:
patch vllm/model_executor/models/llama.py -i file.patch