Bring back CLI image commands ("az ml image create" etc)
The earlier preview versions of the CLI extension (e.g. 1.0.10) contained commands for manipulating the Docker images (e.g. "az ml image create container"). This was very useful, because it made it possible to decouple the time consuming task of building the image from the actual deployment to a compute target and allowed verification of the image prior to deploying to production. (In a CI/CD process, image building was part of the "Build" pipeline while webservice deployment was performed in the "Release" pipeline.)
In the latest CLI extension version (188.8.131.52), the image commands are not present in the CLI, even though the relevant classes (Image, ContainerImage) are still there in the SDK. I understand that some users may want to deploy models directly and not worry about images, but please take into account the needs of those of us who prefer to build the image independently from webservice deployment and bring back the "az ml image" commands to the CLI.
Hi Azure Customer,
This request is now added to the roadmap and in progress. Please keep eyes on it. Thank you for your feedback again.
Azure CXP Community
Gopal Vashishtha (MSFT) commented
Customers should be able to deploy an image built from az ml model package using the below inference configuration, without rebuilding the image every time. Please let me know if this doesn't work:
Dean Van Asseldonk commented
What is the timeline on introducing the "az ml model package" command? We are currently deciding whether we should refactor our build pipeline to use the "az ml model deploy" method or to wait for "az ml model package". We currently use the old "az ml image create" commands.
Jordan Edwards commented
Images are being removed as a top level concept in Azure Machine Learning.
There are a couple of things which we are working on which will improve your experience:
1. We will be introducing a "az ml model package" command which provides the equivalent functionality to what you had in "image create", if required.
2. We are building in model data mounting and base image caching to our platform which should dramatically speed up deployment times for your ML models.
We will reply here once model package is available.