Exam DP-100 topic 4 question 27 discussion

The given answer is correct. A compute target such as AKS needs to be created in order to do inferencing real-time.

A is the correct answer

D. Create an Azure Machine Learning compute cluster.

Answer is D, A has nothing to do before deploy

D is correct, because it says about deployment, this means we have think about deployment only not about whether person has run the inference-pipeline for testing

A for me. "You use the Azure Machine Learning designer to create and run a training pipeline" means you have run training pipeline but it doesn't mean you have run Inferencing Pipeline. You must run inferencing pipeline at least once before deployment.

A is the one

To deploy a real-time inference pipeline, you have to run the pipeline at least once (submit before you deploy it) https://learn.microsoft.com/en-us/azure/machine-learning/tutorial-designer-automobile-price-deploy?view=azureml-api-1

D. Create an Azure Machine Learning compute cluster. In order to deploy a real-time inference pipeline as a web service in Azure Machine Learning, you need to have a compute target, and this can be an Azure Machine Learning Compute Cluster. This is where your model will be hosted and run when you deploy it. The other options are not necessary before deploying the inference pipeline. Running the pipeline or creating a batch inference pipeline isn't required before deployment. Cloning the training pipeline also has no direct bearing on deploying the real-time inference pipeline.

"You use the Azure Machine Learning designer to create and run a training pipeline. You then create a real-time inference pipeline". This means that you already ran the training pipeline, because otherwise you would not have been able to create a real-time inferencing pipeline. So the only logical in my view would be D.

I think the given answer is correct

Exactly this question was on exam 07/March/2023

D. Create an Azure Machine Learning compute cluster.

AKS is not given as one of the options, so i'd vote for A

On exam 18-06-22

A is correct! You must have a target already, so you can train a pipeline, which is a given

Correct answer is D. Read the link provided. After creating the real time inference cluster, a compute cluster must be created to run the inference.