With Custom Containers in Managed Online Endpoints, you can deploy a custom docker container as a managed online endpoint, leverage all the scalability, monitoring, and alerting capabilities of online endpoints but use a custom inferencing stack like TorchServe, TensorFlow Serving, R, or ML.NET. Simply specify the port and path to use for liveness, readiness, and scoring, and we will take care of deploying your custom container as a managed online endpoint.
Job Creation UI on Azure Machine Learning Studio provides a new and consistent job creation experience. You can now use the Studio UI to create and manage training jobs. Using an easy-to-follow wizard, specify your compute, environment, code, data configurations and Azure ML will create the training jobs.
If you see a failed job and want to change some parameters, you would also be able to make some changes to a previous job setting and resubmit your job from the creation flow.
Read the documentation.
Learn about TensorFlow Serving.
Learn about Touchserve.