Azure DP-100 Exam: Configuring MLFlow for ML Workspace

Configuring MLFlow for ML Workspace

Prev Question Next Question

Question

While running ML experiments, you want to use MLFlow to track execution of the runs, to monitor the model training process and to store the metrics of the runs in your ML workspace.

Which of the following steps is not necessary while configuring MLFlow for your ML workspace?

Answers

Explanations

Click on the arrows to vote for the correct answer

A. B. C. D.

Correct Answer: D.

Option A is incorrect because the azureml-mlflow package is a prerequisite to using the MLFlow for monitoring your experiments.

Option B is incorrect because MLFlow's set_tracking_uri is required for MLFlow to point to the URI of an ML workspace.

Option C is incorrect because the workspace's get_mlflow_tracking_uri is used to achieve a unique tracking URI to the workspace (to be used with the set_tracking_uri() later)

Option D is CORRECT because the azureml-core package is automatically installed together with the azureml-mlflow package.

Diagram:

MLflow with Azure Machine Learning Experimentation Experiments Local machine Virtual machine Azure Machine Learning Compute Azure Databricks Experiments and Metrics Logging LE» ml flow Logging API Tracking URI Experiments and Metrics Tracking Azure Machine Learning Workspace E] Metrics Artifacts

Reference:

To configure MLFlow for your ML workspace, you need to perform the following steps:

A. Install the azureml-mlflow package: This package enables integration between Azure Machine Learning and MLFlow. You can install this package using the pip command: !pip install azureml-mlflow.

B. Set the tracking URI: After installing the package, you need to set the tracking URI to connect to your ML workspace. You can set the tracking URI using the following code:

python
from azureml.core import Workspace import mlflow.azureml  # Get the workspace ws = Workspace.from_config()  # Set the tracking URI mlflow.set_tracking_uri(ws.get_mlflow_tracking_uri())

C. Get the MLFlow tracking URI: This step is not necessary while configuring MLFlow for your ML workspace. However, you can use the get_mlflow_tracking_uri method to get the MLFlow tracking URI for your workspace.

D. Install azureml-core: This package is required to work with Azure Machine Learning. However, it is not directly related to configuring MLFlow for your workspace. You can install this package using the pip command: !pip install azureml-core.

Therefore, the step that is not necessary while configuring MLFlow for your ML workspace is C. Get the MLFlow tracking URI.

Prev Question Next Question