Databricks Certified Machine Learning - Associate
Get started today
Ultimate access to all questions.
A data scientist is using MLflow to track a machine learning experiment, including hyperparameter tuning. They aim to organize the experiment with one parent run for the tuning process and child runs for each unique hyperparameter combination, all initiated manually with mlflow.start_run(). What is the best method to achieve this organization?
A data scientist is using MLflow to track a machine learning experiment, including hyperparameter tuning. They aim to organize the experiment with one parent run for the tuning process and child runs for each unique hyperparameter combination, all initiated manually with mlflow.start_run(). What is the best method to achieve this organization?
Real Exam
Explanation:
MLflow supports hierarchical run organization, allowing parent and child runs to be logically grouped. For hyperparameter tuning, specifying nested=True when starting each child run ensures it is correctly nested under the parent run. This method provides a clear structure for reviewing and comparing different hyperparameter combinations.
- Option A is incorrect because Autologging does not organize runs hierarchically.
- Option B is misleading as it lacks the
nested=Trueparameter, which is essential for creating a parent-child relationship. - Option D is incorrect because using the same experiment ID does not establish a hierarchical relationship between runs.
- Option E is incorrect because
nested=Trueshould be specified for child runs, not the parent run.