You have successfully trained your ML model and haven't yet registered it in your workspace.
Now you need to deploy it to the runtime environment as a real-time inference service.
Azure ML offers several methods for deploying a model and you have to choose that best fits your scenario.
Which of the following methods can you use in this case?
Click on the arrows to vote for the correct answer
A. B. C. D.Answer: D.
Option A is incorrect because while the deploy method of the Model is the easiest way to deploy a model to its runtime environment, it can only be used for registered models only.
It would require a registration step so that it can be used.
Option B is incorrect because the deploy_from_model method of the Webservice object can also be used to deploy a model as a webservice, however it does not register the model.
Use it for models already registered in the workspace.
Option C is incorrect because the deploy_from_image method of the Webservice object can be used if you already have an image to be deployed.
It takes an Image as an input parameter instead of a Model.
Since you have finished with the training of your model, you probably don't have an image yet.
Option D is CORRECT because the deploy method of the Webservice object can be used to deploy a model even if it is not registered.
The method completes both its registration and deployment as a real-time service.
This is the solution for deploying unregistered models in one step.
Reference:
In this scenario, to deploy a trained ML model as a real-time inference service in Azure, you can use the "deploy method of the Webservice" (option D).
Option A, "deploy method of the Model," is not correct as it does not allow deploying the model as a web service.
Option B, "deploy_from_model method of the Webservice," is not the ideal choice for this scenario as it is used when a model has already been registered in the workspace and needs to be deployed. In this case, the model has not been registered yet.
Option C, "deploy_from_image method of the Webservice," is used to deploy an image of a pre-configured web service deployment environment. This is not relevant in this scenario as we need to deploy the trained ML model as a real-time inference service.
Therefore, option D, "deploy method of the Webservice," is the correct choice. This method deploys the model as a real-time inference service and automatically registers the model in the workspace. Additionally, this method provides several deployment options such as ACI (Azure Container Instances), AKS (Azure Kubernetes Service), and Azure Functions.
To use this method, you first need to create a deployment configuration that specifies the type of deployment environment, the compute target, and the number of instances. Then, you can call the "deploy" method of the Webservice object, passing in the deployment configuration and the trained model object. After the deployment is complete, you can consume the deployed service by sending HTTP requests to its endpoint.