Google Professional Machine Learning Engineer Question 438
Select 3Google Cloud PlatformYou are a machine learning engineer working at a retail company. You have developed multiple versions of a demand forecasting model. To track and compare the performance of each model version, you decide to use Vertex AI Experiments. Which of the following steps should you take to effectively track and compare your model artifacts and versions in Vertex AI Experiments?
- A
Log experiment parameters, metrics, and model artifacts using the Vertex AI SDK.
- B
Manually create a spreadsheet to record model version details and performance metrics.
- C
Use Vertex AI Pipelines to automate the execution of experiments and log experiment metadata.
- D
Visualize and compare experiment runs using the Vertex AI Experiments dashboard.
- E
Store model artifacts in a local directory for quick access during comparisons.
Show answer and explanation
Correct answers: A, C, D
Explanation
To effectively track and compare model artifacts and versions in Vertex AI Experiments, you should use tools and features provided by the Vertex AI platform. Logging experiment metadata with the Vertex AI SDK and automating this process with Vertex AI Pipelines ensures consistency and scalability. The Vertex AI Experiments dashboard provides a visual interface for easy comparison of different experiment runs. Manual methods or local storage are not aligned with best practices for experiment tracking in the cloud.
- A. Correct.
Correct. Logging experiment parameters, metrics, and model artifacts using the Vertex AI SDK allows you to track and store experiment details in a structured and scalable way.
- B. Incorrect.
Incorrect. Using a manual spreadsheet is error-prone and does not leverage Vertex AI’s built-in tools for experiment tracking.
- C. Correct.
Correct. Vertex AI Pipelines can automate the execution of experiments and ensure that metadata, such as parameters and metrics, are logged consistently.
- D. Correct.
Correct. The Vertex AI Experiments dashboard provides a visual interface to compare experiment runs, making it easier to analyze model performance.
- E. Incorrect.
Incorrect. Storing model artifacts in a local directory is not recommended for tracking and comparing experiments effectively, as it lacks integration with Vertex AI’s experiment tracking tools.