Databricks Machine Learning Professional Question 82
Select 3A data scientist is working on a machine learning model in Databricks and has registered the model in the MLflow Model Registry. They want to add metadata to both the registered model and one of its specific versions. Which of the following actions can they perform to achieve this?
- A
Use the MLflow API to add tags to the registered model.
- B
Add a description to the registered model version using the Databricks UI.
- C
Update the model version stage to 'Production' to include metadata.
- D
Use the MLflow API to add tags to the registered model version.
- E
Directly edit the model's artifact files to include metadata.
Show answer and explanation
Correct answers: A, B, D
Explanation
In Databricks, metadata for registered models and their versions can be managed using MLflow. Tags can be added programmatically using the MLflow API for both registered models and their versions. Additionally, the Databricks UI allows users to add descriptions to registered model versions. However, changing a model stage or editing artifact files does not serve as a correct approach to manage metadata.
- A. Correct.
This is correct. The MLflow API provides functionality to add tags to registered models, which can be used as metadata.
- B. Correct.
This is correct. The Databricks UI allows users to add descriptions to specific model versions as metadata.
- C. Incorrect.
This is incorrect. Updating the model version stage (e.g., to 'Production') does not inherently add metadata but changes the model's lifecycle stage.
- D. Correct.
This is correct. The MLflow API can be used to add tags to specific model versions, enabling the addition of metadata at the version level.
- E. Incorrect.
This is incorrect. Editing the model's artifact files is not a supported or recommended method to add metadata in Databricks.