Databricks Machine Learning Professional Question 96
Select 2You are working on a machine learning project in Databricks, and you want to automate the process of transitioning a model to 'Production' only after it passes all the validation tests. Which of the following steps should you take to achieve this using Model Registry Webhooks and Databricks Jobs?
- A
Configure a Model Registry Webhook to trigger a Databricks Job whenever the model's stage is updated to 'Staging'.
- B
Create a Databricks Job that runs validation tests against the model version and updates the model's stage to 'Production' if the tests pass.
- C
Manually trigger the Databricks Job from the Model Registry whenever a new model version is registered.
- D
Use a Model Registry Webhook to automatically delete a model version if its validation tests fail.
- E
Set up a Model Registry Webhook to notify stakeholders via email after a model transitions to 'Production'.
Show answer and explanation
Correct answers: A, B
Explanation
To automate the model lifecycle in Databricks, you can use Model Registry Webhooks to trigger specific actions when events occur, such as when a model stage changes. By configuring a webhook to trigger a Databricks Job on a stage update and ensuring the job performs validation tests to decide if the model is ready for 'Production', you create a seamless, automated workflow. Other actions, such as manual triggers or email notifications, are not directly relevant to automation and are outside the scope of this scenario.
- A. Correct.
Correct: Configuring a Model Registry Webhook to trigger a Databricks Job on a stage update ensures that the validation process is automated whenever a model is promoted to 'Staging'.
- B. Correct.
Correct: This is a necessary step to automate the transition of a model to 'Production' based on the results of the validation tests.
- C. Incorrect.
Incorrect: Manually triggering the job defeats the purpose of automation, which is the focus of this question.
- D. Incorrect.
Incorrect: While deleting a model version could be part of a workflow, it is not directly relevant to automating the promotion of a model to 'Production'.
- E. Incorrect.
Incorrect: Sending notifications to stakeholders is not directly related to automating the model lifecycle using Model Registry Webhooks and Databricks Jobs.