Cannot deploy Phi-3-mini-4k-instruct model on AWS Sagemaker

#51
by MajTad - opened

I try to run a sagemaker endpoint with the Phi-3-mini-4k-instruct using HuggingFaceModel to load it but I got from Sagemaker logs the following error:

ValueError: Unsupported model type phi3.

Does anyone face the same problem.

It's only made to run on Azure
/s

Seriously, not all models are supported by the HuggingFaceModel feature on SageMaker. You can just deploy it on a regular gpu-powered vm using vLLM or TGI

Thanks @amgadsahl for your answer.
I'll manage to deploy based on your recommandations.

nguyenbh changed discussion status to closed

Sign up or log in to comment