Question 8
Domain 1: Design and prepare a machine learning solutionYou need to run a script that trains a deep neural network (DNN) model and logs the loss and accuracy metrics. You decide to run the training script as an experiment on the aks-cluster compute target. Will this solution work properly?
Correct answer: B
Explanation
No. An AKS cluster is designed for serving models and inference, while training experiments require a supported training compute target such as an Azure Machine Learning compute cluster. Since the script “trains a deep neural network (DNN) model and logs the loss and accuracy metrics,” running it on aks-cluster will not meet the intended use of that target.
Why each option is right or wrong
A. Yes, because an AKS cluster can be used as a compute target for running training experiments and logging metrics.
B. No, because an AKS cluster is intended for deployment and inference workloads, not as a supported compute target for training experiments.
Azure Kubernetes Service is not a supported compute target for Azure Machine Learning training jobs; it is used for deployment and online inference, while training experiments must run on supported training targets such as an Azure ML compute cluster. Because the script is explicitly training a DNN and logging metrics, submitting it to the aks-cluster target would be the wrong workload type and therefore would not work as intended.