You use the Azure Machine Learning designer to create and run a training pipeline. You then create a real-time inference pipeline. You must deploy the real-time inference pipeline as a web service…

The correct answer is D. Create an Azure Machine Learning compute cluster. Before deploying a real-time inference pipeline as a web service, an inference compute target must exist to host the deployed endpoint. For real-time web services in Azure ML, this is an AKS-based inference cluster. Without a suitable compute target, the Deploy option is…

Train and deploy models

Question

You use the Azure Machine Learning designer to create and run a training pipeline. You then create a real-time inference pipeline. You must deploy the real-time inference pipeline as a web service. What must you do before you deploy the real-time inference pipeline?

Options

ARun the real-time inference pipeline.
BCreate a batch inference pipeline.
CClone the training pipeline.
DCreate an Azure Machine Learning compute cluster.

How the community answered

(37 responses)

A
8% (3)
B
3% (1)
C
3% (1)
D
86% (32)

Explanation

Before deploying a real-time inference pipeline as a web service, an inference compute target must exist to host the deployed endpoint. For real-time web services in Azure ML, this is an AKS-based inference cluster. Without a suitable compute target, the Deploy option is unavailable. Running the inference pipeline (option A) is also a prerequisite-the pipeline must be run successfully at least once before the Deploy button becomes active-but the question identifies the compute infrastructure as the key prerequisite to create. Creating a batch inference pipeline or cloning the training pipeline are not required steps.

Topics

#Azure ML deployment#Real-time inference#Compute cluster#Web service

Community Discussion

No community discussion yet for this question.

Full DP-100 Practice