Azure Data Scientists Associate Practice Exam

Which method should be used to submit a training script for a machine learning model in Azure Machine Learning?

1. Run.get.register()

2. Model.register()

3. Experiment.submit()

4. Model.download()

The correct answer is: Experiment.submit()

The method used to submit a training script for a machine learning model in Azure Machine Learning is to utilize the Experiment.submit() function. This method is designed specifically for submitting experiments, which include the execution of training scripts. When you call this method, you provide it with an Estimator, which contains the specifics of the training environment, script, and other necessary parameters for model training. The Experiment class in Azure Machine Learning allows you to manage and track different iterations of your experiments, making it an essential part of the development workflow. By submitting your training script through this method, you ensure that all relevant metadata about the run—like metrics, outputs, and logs—is captured, giving you the ability to monitor and evaluate your model's performance effectively. In contrast, the other options serve different purposes within Azure Machine Learning. For instance, methods like Run.get.register() and Model.register() focus on registering already-trained models or components, while Model.download() is intended for retrieving a trained model from the workspace rather than submitting training scripts. This distinction highlights why Experiment.submit() is the appropriate choice for initiating training processes in Azure Machine Learning.