A retail company stores customer purchase history in BigQuery and product images in Cloud Storage. They want to train a custom TensorFlow model that combines tabular purchase features with image embeddings. The training data is large (several TB) and the team wants to minimize data movement. Which approach should they use?

Use the BigQuery Storage API directly from the custom training job to read the tabular data, and read images from Cloud Storage via the file system.

Load the images into BigQuery as BLOB data and train the model using a single BigQuery dataset.

Export the BigQuery table to CSV files in Cloud Storage, then have the training job read both the CSV and image files from Cloud Storage.

Copy the entire BigQuery table to a local data center and then upload it to Cloud Storage for training.

A team is using Kubeflow on GKE to run a distributed training job. The job is failing because the worker pods cannot connect to the chief pod. The team has verified that the pods are running and the service names are correct. What is the most likely cause?

The training code is incorrect.

The network policy is blocking traffic between pods.

The GKE cluster does not have enough resources.

The pods are in different Kubernetes namespaces.

Google CloudProfessional Machine Learning Engineer

Domain 3Objective 2

Training Models PROFESSIONAL-MACHINE-LEARNING-ENGINEER Practice Questions (Page 6)

Part of the Scaling prototypes into ML models domain, which accounts for ~21% of the PROFESSIONAL-MACHINE-LEARNING-ENGINEER exam.

31questions here

7free pages

8concepts

~21%of the exam

Questions 26–30

26foundation · easy
When submitting a training job to Agent Platform, which of the following can be specified to control the compute resources allocated to the job?
Select an answer first
27application · medium
A retail company stores customer purchase history in BigQuery and product images in Cloud Storage. They want to train a custom TensorFlow model that combines tabular purchase features with image embeddings. The training data is large (several TB) and the team wants to minimize data movement. Which approach should they use?
Select an answer first
28expert · medium
A team is using Kubeflow on GKE to run a distributed training job. The job is failing because the worker pods cannot connect to the chief pod. The team has verified that the pods are running and the service names are correct. What is the most likely cause?
Select an answer first
29foundation · easy
A data scientist needs to move a large batch of historical CSV files from an on-premises server to Cloud Storage for use in a training pipeline. Which Google Cloud service is designed for this type of batch transfer?
Select an answer first
30foundation · easy
A data scientist with limited ML coding experience wants to train a custom image classification model without writing any training code. Which Google Cloud service is most appropriate?
Select an answer first

Finished these 5 questions?

Review the revealed explanations, or continue through the curriculum.

Next 5 questions →Back to PROFESSIONAL-MACHINE-LEARNING-ENGINEER

Free Basic Practice is a study aid with revealable answers — not a scored exam. Examers.io is independent and not affiliated with or endorsed by Google Cloud. “PROFESSIONAL-MACHINE-LEARNING-ENGINEER” is a trademark of its owner, used for identification only.

Training Models PROFESSIONAL-MACHINE-LEARNING-ENGINEER Practice Questions (Page 6)

Questions 26–30

When submitting a training job to Agent Platform, which of the following can be specified to control the compute resources allocated to the job?

A team is using Kubeflow on GKE to run a distributed training job. The job is failing because the worker pods cannot connect to the chief pod. The team has verified that the pods are running and the service names are correct. What is the most likely cause?

A data scientist needs to move a large batch of historical CSV files from an on-premises server to Cloud Storage for use in a training pipeline. Which Google Cloud service is designed for this type of batch transfer?

A data scientist with limited ML coding experience wants to train a custom image classification model without writing any training code. Which Google Cloud service is most appropriate?

Keep studying — Scaling prototypes into ML models