Databricks Certified Generative AI Engineer Associate Exam - Topic 2 Question 5 Discussion

Actual exam question for Databricks's Databricks Certified Generative AI Engineer Associate exam

Question #: 5
Topic #: 2

[All Databricks Certified Generative AI Engineer Associate Questions]

A Generative Al Engineer has already trained an LLM on Databricks and it is now ready to be deployed.

Which of the following steps correctly outlines the easiest process for deploying a model on Databricks?

ALog the model as a pickle object, upload the object to Unity Catalog Volume, register it to Unity Catalog using MLflow, and start a serving endpoint

BLog the model using MLflow during training, directly register the model to Unity Catalog using the MLflow API, and start a serving endpoint

CSave the model along with its dependencies in a local directory, build the Docker image, and run the Docker container

DWrap the LLM's prediction function into a Flask application and serve using Gunicorn

Show Suggested Answer

Suggested Answer: B

Problem Context: The goal is to deploy a trained LLM on Databricks in the simplest and most integrated manner.

Explanation of Options:

Option A: This method involves unnecessary steps like logging the model as a pickle object, which is not the most efficient path in a Databricks environment.

Option B: Logging the model with MLflow during training and then using MLflow's API to register and start serving the model is straightforward and leverages Databricks' built-in functionalities for seamless model deployment.

Option C: Building and running a Docker container is a complex and less integrated approach within the Databricks ecosystem.

Option D: Using Flask and Gunicorn is a more manual approach and less integrated compared to the native capabilities of Databricks and MLflow.

Option B provides the most straightforward and efficient process, utilizing Databricks' ecosystem to its full advantage for deploying models.

by Doretha at Oct 23, 2024, 04:30 AM

Limited Time Offer

25%

Off

Get Premium Databricks Certified Generative AI Engineer Associate Questions as Interactive Web-Based Practice Test or PDF

Contribute your Thoughts:

Submit Cancel

Rolland

3 months ago

D is interesting, but isn't it a bit outdated for model serving?

upvoted 0 times

...

Lajuana

3 months ago

Wait, can you really just use MLflow like that? Sounds too easy!

upvoted 0 times

...

Barney

3 months ago

C seems way too complicated for deployment.

upvoted 0 times

...

Serita

4 months ago

I think A is more straightforward, though.

upvoted 0 times

...

Nohemi

4 months ago

Option B is definitely the easiest way to go!

upvoted 0 times

...

Nidia

4 months ago

I don't think wrapping the function in Flask is the easiest method for deployment on Databricks, but it could work in other contexts.

upvoted 0 times

...

Dahlia

4 months ago

I feel like saving the model as a pickle object was mentioned, but I can't recall if that's the best way to deploy on Databricks.

upvoted 0 times

...

Sylvie

4 months ago

I think option B sounds familiar because it mentions using the MLflow API, which we practiced in class.

upvoted 0 times

...

Levi

5 months ago

I remember we discussed using MLflow for model logging, but I'm not sure if it was specifically about registering to Unity Catalog.

upvoted 0 times

...

Jovita

5 months ago

Ah, this is a good one. I remember learning about Databricks deployment in class. I think option B is the way to go - it's the most direct approach and aligns with what we covered.

upvoted 0 times

...

Suzi

5 months ago

Option A looks promising, but I'm not sure about the Unity Catalog part. I'll need to review my notes on Databricks deployment to make sure I understand that properly.

upvoted 0 times

...

Elke

5 months ago

Hmm, I'm a bit unsure about this one. The options all seem to involve different steps, and I'm not entirely familiar with the Databricks deployment process. I'll need to think this through carefully.

upvoted 0 times

...

Kristian

5 months ago

This seems like a straightforward deployment question. I think I'll go with option B - it looks like the easiest way to get the model registered and deployed on Databricks.

upvoted 0 times

...