latest vllm docker (v0.6.2) fail to load

#1
by choronz333 - opened

--model=neuralmagic/Phi-3.5-mini-instruct-FP8-KV

OSError: neuralmagic/Phi-3.5-mini-instruct-FP8-KV does not appear to have a file named configuration_phi3.py. Checkout 'https://huggingface.co/neuralmagic/Phi-3.5-mini-instruct-FP8-KV/tree/main' for available files.

choronz333 changed discussion title from vllm docker load fail to load to latest vllm docker (v0.6.2) fail to load

Yes, can confirm, seems like the problem comes from the OpenAI server. The offline code works.

UPD after trying and failing to start the OpenAI instance the offline code stopped working

Neural Magic org

Thanks for reporting this, I have added the config and modeling files.

mgoin changed discussion status to closed

Sign up or log in to comment