Is it possible to take a ML Azure Container Instance from the registry and deploy it elsewhere?

Philip Brown 6 Reputation points
2021-01-13T16:21:42.053+00:00

Hi,

I was wondering if it was possible to take a ML Model deployed to a container instance and then take that and deploy that in docker elsewhere. I have managed to sync a repo with the Azure registry and download the container which was created in Azure ML. When i started the container though it didn't do anything; i logged into the container and nothing was running; i could start NGINX but wasn't sure if this was a supported?

I was that you can do this with the Azure Sentiment Analysis Containers but wondering if you could do with ML Azure Container Instances?

Any pointers would be good.

Azure Container Instances
Azure Container Instances
An Azure service that provides customers with a serverless container experience.
670 questions
Azure Machine Learning
Azure Machine Learning
An Azure machine learning service for building and deploying models.
2,689 questions

1 answer

Sort by: Most helpful
Most helpful Newest Oldest
  1. GiftA-MSFT 11,161 Reputation points
    2021-01-15T18:26:56.77+00:00

    Thanks for reaching out. Yes, however, you need to run az ml model package to get an image with your model baked into it, otherwise, the model file is mounted by the deployment engine. Here's more information on how to package a registered model with Docker. Hope this helps.

    1 person found this answer helpful.
    0 comments