Google Professional Machine Learning Engineer Question 443
Select 3Google Cloud PlatformYou are building a machine learning pipeline in Google Cloud Vertex AI and want to track lineage, artifacts, and events for better reproducibility and debugging. Which of the following actions can you take using Vertex ML Metadata to achieve this?
- A
Track input and output datasets used in your training jobs.
- B
Log hyperparameters and evaluation metrics for your models.
- C
Automatically deploy models to Vertex AI endpoints.
- D
Visualize the lineage of your artifacts, such as datasets and models.
- E
Store and retrieve model predictions for real-time serving.
Show answer and explanation
Correct answers: A, B, D
Explanation
Vertex ML Metadata is a tool for tracking and managing metadata related to machine learning workflows. It supports tracking datasets, hyperparameters, evaluation metrics, and artifact lineage, which are crucial for reproducibility and debugging. However, it does not handle tasks like model deployment or serving predictions, which are outside its scope.
- A. Correct.
Vertex ML Metadata allows you to track the lineage of datasets, including input and output datasets, as part of its artifact tracking functionality.
- B. Correct.
Vertex ML Metadata supports logging hyperparameters and evaluation metrics as part of the metadata stored for machine learning experiments.
- C. Incorrect.
While Vertex AI supports model deployment, this is not a feature of Vertex ML Metadata. Metadata focuses on tracking and lineage, not deployment.
- D. Correct.
Vertex ML Metadata provides tools to visualize the lineage of your artifacts, which helps in understanding and debugging ML workflows.
- E. Incorrect.
Vertex ML Metadata does not store or retrieve model predictions. This is typically managed through other services like Vertex AI Prediction.