Databricks Machine Learning Associate Question 466
Select 3You are a data scientist using Databricks and have trained multiple models for a project. You want to compare model performance and understand which hyperparameters were used. Which of the following types of information can you access in the MLflow UI for a specific experiment?
- A
Model performance metrics such as accuracy or RMSE
- B
The source code version used to run the experiment
- C
The name of the person who approved the model for deployment
- D
Hyperparameter values used for training the model
- E
Execution logs and detailed error stack traces of the experiment
Show answer and explanation
Correct answers: A, B, D
Explanation
The MLflow UI is designed to help users track and compare machine learning experiment runs. It provides key details such as logged metrics, hyperparameters, and source code version, which are essential for reproducibility and performance analysis. However, it does not include information related to model approval workflows or detailed execution logs.
- A. Correct.
Correct: The MLflow UI allows users to log and visualize performance metrics such as accuracy, loss, RMSE, etc., for each experiment run.
- B. Correct.
Correct: MLflow can log the version of source code (e.g., Git commit hash) associated with a specific experiment run, and this information is accessible in the UI.
- C. Incorrect.
Incorrect: While the MLflow UI tracks metadata for experiment runs, it does not store information about model approval processes or who approved the model.
- D. Correct.
Correct: The MLflow UI displays hyperparameter values logged during training, making it easy to compare different runs.
- E. Incorrect.
Incorrect: MLflow UI does not provide detailed execution logs or error stack traces. Such logs are generally available in the Databricks workspace's cluster event logs or Spark logs.