Google Professional Machine Learning Engineer Question 437
Single answerGoogle Cloud PlatformYou are a machine learning engineer working on a model that predicts user purchasing behavior. You have iteratively trained multiple versions of the model with different hyperparameter configurations. You want to track and compare the performance metrics and configurations of these models to identify the best-performing version. How can you achieve this using Google Cloud's Vertex AI?
- A
Use Vertex AI Experiments to log and compare model metrics and configurations across training runs.
- B
Store all model artifacts in a Cloud Storage bucket and manually track the configurations and metrics in a spreadsheet.
- C
Use Vertex AI Model Registry to log and compare model metrics and configurations during training.
- D
Create a custom database in BigQuery to store and query the training metrics and configurations for comparison.
Show answer and explanation
Correct answer: A
Explanation
Vertex AI Experiments is specifically designed to track and compare model training runs, including their configurations and performance metrics. It provides a streamlined and automated way to manage this process, avoiding the manual effort or custom development required by other options.
- A. Correct.
This is correct. Vertex AI Experiments allows you to log, track, and compare model training runs, including metrics and configurations, making it the recommended approach for this scenario.
- B. Incorrect.
This is incorrect. Manually tracking model metrics and configurations is error-prone and inefficient, especially when scalable tools like Vertex AI Experiments are available.
- C. Incorrect.
This is incorrect. Vertex AI Model Registry is used for managing deployed models and their versions, not for tracking training runs or comparing metrics.
- D. Incorrect.
This is incorrect. While storing data in BigQuery is possible, it requires significant custom development and does not leverage the built-in capabilities of Vertex AI Experiments.