nerdexam
Google

PROFESSIONAL-MACHINE-LEARNING-ENGINEER · Question #253

Your team is training a large number of ML models that use different algorithms, parameters, and datasets. Some models are trained in Vertex AI Pipelines, and some are trained on Vertex AI Workbench…

The correct answer is B. Create a Vertex AI experiment. Submit all the pipelines as experiment runs. For models trained. To compare models trained across Vertex AI Pipelines and Workbench with minimal effort, create a Vertex AI experiment, submit pipeline runs as experiment runs, and log metrics for Workbench models to the same experiment.

Submitted by rohit_dlh· Apr 18, 2026ML model development

Question

Your team is training a large number of ML models that use different algorithms, parameters, and datasets. Some models are trained in Vertex AI Pipelines, and some are trained on Vertex AI Workbench notebook instances. Your team wants to compare the performance of the models across both services. You want to minimize the effort required to store the parameters and metrics. What should you do?

Options

  • AImplement an additional step for all the models running in pipelines and notebooks to export
  • BCreate a Vertex AI experiment. Submit all the pipelines as experiment runs. For models trained
  • CImplement all models in Vertex AI Pipelines Create a Vertex AI experiment, and associate all
  • DStore all model parameters and metrics as model metadata by using the Vertex AI Metadata API.

How the community answered

(23 responses)
  • A
    4% (1)
  • B
    74% (17)
  • C
    13% (3)
  • D
    9% (2)

Why each option

To compare models trained across Vertex AI Pipelines and Workbench with minimal effort, create a Vertex AI experiment, submit pipeline runs as experiment runs, and log metrics for Workbench models to the same experiment.

AImplement an additional step for all the models running in pipelines and notebooks to export

Exporting metrics to CSV files and manually combining them creates significant overhead for managing and comparing a large number of models and runs.

BCreate a Vertex AI experiment. Submit all the pipelines as experiment runs. For models trainedCorrect

Vertex AI Experiments is explicitly designed for tracking and comparing ML experiments, regardless of where they are executed. By submitting pipeline runs as experiment runs and programmatically logging metrics for Workbench models to the same experiment, you centralize all parameters and metrics for easy comparison through the Vertex AI console, minimizing manual effort.

CImplement all models in Vertex AI Pipelines Create a Vertex AI experiment, and associate all

Migrating all models to Vertex AI Pipelines would require substantial refactoring effort, contradicting the goal of minimizing effort, especially when Workbench is already in use for some models.

DStore all model parameters and metrics as model metadata by using the Vertex AI Metadata API.

Storing parameters and metrics using the Vertex AI Metadata API is part of the solution, but it doesn't inherently provide the comparison and visualization capabilities of Vertex AI Experiments; you would still need to build custom tools on top of it.

Concept tested: Vertex AI Experiments for unified model comparison

Source: https://cloud.google.com/vertex-ai/docs/experiments/overview

Topics

#Vertex AI Experiments#Model tracking#Performance comparison#MLOps

Community Discussion

No community discussion yet for this question.

Sign up to join the discussion
Full PROFESSIONAL-MACHINE-LEARNING-ENGINEER Practice