Google Professional Machine Learning Engineer Question 447
Select 4Google Cloud PlatformYou are managing an end-to-end machine learning pipeline on Google Cloud. To ensure reproducibility and track changes effectively, you need to implement model and dataset versioning. Which of the following steps should you take to achieve this?
- A
Use Vertex AI Model Registry to store and version models.
- B
Store datasets in Cloud Storage with versioning enabled.
- C
Use BigQuery to store datasets and rely on table snapshots for version control.
- D
Manually document model changes in a spreadsheet.
- E
Integrate a source control system like Git to track changes in model code and configurations.
Show answer and explanation
Correct answers: A, B, C, E
Explanation
To ensure proper model and dataset versioning in a machine learning workflow on Google Cloud, you should use tools designed for this purpose. Vertex AI Model Registry is ideal for model versioning, and Cloud Storage with versioning or BigQuery table snapshots provides effective dataset versioning. Additionally, integrating Git for tracking code and configuration changes is a standard practice. Manual approaches, like using spreadsheets, are unreliable and not recommended for production-grade workflows.
- A. Correct.
Vertex AI Model Registry is specifically designed for storing and versioning machine learning models, making it an essential choice for model management.
- B. Correct.
Cloud Storage provides a native versioning feature that helps track changes to datasets, ensuring a history of modifications.
- C. Correct.
BigQuery table snapshots allow you to create point-in-time copies of your data, which is useful for dataset versioning.
- D. Incorrect.
Manually documenting model changes in a spreadsheet is error-prone and not scalable, making it an unsuitable approach for version control.
- E. Correct.
A source control system like Git is a best practice for tracking changes in model code and configurations, ensuring proper versioning and reproducibility.