AI-300 Practice Question 85: MLflow Tracking URI Local

Microsoft Machine Learning Operations Engineer Associate (AI-300) · Test 9 · question 5 of 10

Implement machine learning model lifecycle and operationsmedium

A developer trains models on a laptop with a script that uses MLflow. The runs are saved only on the laptop, but the team wants the metrics and parameters to appear in the Azure Machine Learning workspace. What should the developer do?

Select one answer
Show answer and explanation

Answer:

  • A. Point MLflow at the workspace's tracking URI by using mlflow.set_tracking_uri

An Azure Machine Learning workspace is an MLflow tracking server. The developer reads the workspace's mlflow_tracking_uri, for example with az ml workspace show --query mlflow_tracking_uri, and passes it to mlflow.set_tracking_uri() (with the azureml-mlflow package installed). Runs logged from the laptop then appear in the workspace (option A).

- B is wrong because the files would sit in storage but not be registered as runs, so they would not appear in the workspace's jobs and metrics. - C is wrong because this only shares a local server and does not send anything to the workspace. - D is wrong because a component defines a reusable pipeline step. It does not redirect the MLflow tracking of a script on a laptop.

Reference: Configure MLflow for Azure Machine Learning

Practice Test 9 ›

Study guides for AI-300