How to use batch endpoints (preview) in Azure Machine Learning studio tutorial

Question

Hello,

I have been following this tutorial on how to use batch endpoints.

When I attempt to select environment I am not greeted with the same as on the tutorial, please see below screenshot.

What are the steps in continuing this tutorial in terms of selecting an environment?

Kind Regards,

Adam N

Answer 1

@Adam Nevin Currently, the scoring and dependencies files are generated only for mlflow models that are registered with Azure ML workspace.
For example after running this notebook locally I registered the model and the folder with Azure ML workspace from the models tab.
This registered model can be selected while creating the batch endpoint and the dependencies are automatically created without the need of selecting a custom environment.

If you are using a different framework model then the dependencies or scoring files need to be provided along with selection of custom environment. The custom environments though first need to be created from the Environments tab by providing a YAML file. After adding a custom environment you can proceed to select the environment on this screen and deploy the endpoint.

Hope this helps!!

If an answer is helpful, please click on or upvote which might help other community members reading this thread.