Google Exam Professional Machine Learning Engineer Topic 4 Question 73 Discussion

Actual exam question for Google's Professional Machine Learning Engineer exam

Question #: 73
Topic #: 4

[All Professional Machine Learning Engineer Questions]

You have a custom job that runs on Vertex Al on a weekly basis The job is Implemented using a proprietary ML workflow that produces the datasets. models, and custom artifacts, and sends them to a Cloud Storage bucket Many different versions of the datasets and models were created Due to compliance requirements, your company needs to track which model was used for making a particular prediction, and needs access to the artifacts for each model. How should you configure your workflows to meet these requirement?

AConfigure a TensorFlow Extended (TFX) ML Metadata database, and use the ML Metadata API.

BCreate a Vertex Al experiment, and enable autologging inside the custom job

CUse the Vertex Al Metadata API inside the custom Job to create context, execution, and artifacts for each model, and use events to link them together.

DRegister each model in Vertex Al Model Registry, and use model labels to store the related dataset and model information.

Show Suggested Answer

Suggested Answer: D

by Tammara at Mar 19, 2024, 03:08 PM

Limited Time Offer

25%

Off

Get Premium Professional Machine Learning Engineer Questions as Interactive Web-Based Practice Test or PDF

Contribute your Thoughts:

Submit Cancel

Lauran

6 months ago

I personally think option C) Use the Vertex AI Metadata API inside the custom Job is the most efficient solution.

upvoted 0 times

...

Gwenn

6 months ago

I disagree, I believe option D) Register each model in Vertex AI Model Registry is the way to go.

upvoted 0 times

...

Brett

7 months ago

I think the best option is A) Configure a TensorFlow Extended (TFX) ML Metadata database, and use the ML Metadata API.

upvoted 0 times

...

Naomi

7 months ago

I'm not sure, but option C sounds good to me. Using Vertex AI Metadata API inside the job seems like a practical approach.

upvoted 0 times

...

Thomasena

7 months ago

I agree with Georgeanna. It's important to have a centralized database to keep track of all the models and datasets.

upvoted 0 times

...

Georgeanna

7 months ago

I think option A is the best choice. By using TFX ML Metadata database, we can easily track the model used for predictions.

upvoted 0 times

...

Twana

8 months ago

Hmm, I'm not sure about option B. Relying on autologging in Vertex AI may not give us enough control over the metadata. And a separate TFX metadata database (option A) sounds like overkill for this use case.

upvoted 0 times

...

Eladia

8 months ago

Option D also sounds promising - registering the models in the Vertex AI Model Registry and using labels could be a simple way to manage the versioning and provenance. But I'm not sure how robust that would be for a complex workflow.

upvoted 0 times

...

Honey

8 months ago

I'm leaning towards option C. Using the Vertex AI Metadata API seems like the most direct way to link the models, datasets, and artifacts together. Plus, we can create custom context and execution details to meet the compliance needs.

upvoted 0 times

Emerson

7 months ago

I'm convinced, option C it is!

upvoted 0 times

...

Shawnee

7 months ago

It's definitely a more structured way to track model usage and artifacts.

upvoted 0 times

...

Brande

8 months ago

Using events to link everything sounds like a good approach.

upvoted 0 times

...

Alona

8 months ago

That makes sense, it's important to link all the necessary information together.

upvoted 0 times

...

Kristeen

8 months ago

We can create custom context and execution details as needed for compliance.

upvoted 0 times

...

Norah

8 months ago

Agreed, using the Vertex AI Metadata API seems like the most direct solution.

upvoted 0 times

...

Truman

8 months ago

I think option C is the best choice.

upvoted 0 times

...

Chaya

8 months ago

Whoa, this question looks like a real brain-teaser! We definitely need to track the models and artifacts for compliance, but it's not clear which option is the best approach.

upvoted 0 times

...