Google Professional Machine Learning Engineer Exam - Topic 2 Question 16 Discussion

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

Question #: 16
Topic #: 2

[All Professional Machine Learning Engineer Questions]

You are training a Resnet model on Al Platform using TPUs to visually categorize types of defects in automobile engines. You capture the training profile using the Cloud TPU profiler plugin and observe that it is highly input-bound. You want to reduce the bottleneck and speed up your model training process. Which modifications should you make to the tf .data dataset?

Choose 2 answers

AUse the interleave option for reading data

BReduce the value of the repeat parameter

CIncrease the buffer size for the shuffle option.

DSet the prefetch option equal to the training batch size

EDecrease the batch size argument in your transformation

Show Suggested Answer

Suggested Answer: A, E

by Jolene at May 07, 2022, 07:29 AM

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

Laurel

4 months ago

I disagree, I think B is actually more useful than people realize!

upvoted 0 times

...

Elena

4 months ago

A and D seem like the most logical picks to me.

upvoted 0 times

...

Lili

4 months ago

Surprised that reducing the repeat parameter (B) is even an option here!

upvoted 0 times

...

Lisandra

4 months ago

I think D could also work, but not sure if it's the best choice.

upvoted 0 times

...

Joesph

5 months ago

Definitely go with A and C! They really help with input bottlenecks.

upvoted 0 times

...

Cecily

5 months ago

I feel like reducing the repeat parameter isn't really going to help with the input bottleneck. It seems more about how we handle the data flow rather than the number of repeats.

upvoted 0 times

...

Melissa

5 months ago

I practiced a similar question, and I think setting the prefetch option to the training batch size could be a good choice. It might help in reducing the waiting time.

upvoted 0 times

...

Iluminada

5 months ago

I'm not entirely sure, but I think increasing the buffer size for shuffling could also help. It seems like it would allow more data to be ready for training.

upvoted 0 times

...

Teri

5 months ago

I remember we discussed using the interleave option to improve data loading speeds. That might help with the input-bound issue.

upvoted 0 times

...

Deeanna

5 months ago

Okay, let me break this down. Test automation is a way to systematically test software, so I think C is the best answer here - it's a strategy, not just a single procedure or objective.

upvoted 0 times

...

Goldie

5 months ago

This one seems pretty straightforward. I'm pretty sure the command to display the open ports is netstat.

upvoted 0 times

...

Mi

5 months ago

I remember a practice question about Cisco FMC settings, and it mentioned disabling management connections. Could that be option D?

upvoted 0 times

...

Justine

5 months ago

I'm a bit confused on this one. I'll need to review the material on ASP.NET deployment before I can confidently answer.

upvoted 0 times

...

Jennie

5 months ago

Looking at the choices, I think $2466 sounds too high for the knee procedure. I'm leaning towards $2305, but I really need to double-check the activity costs.

upvoted 0 times

...