Deal of The Day! Hurry Up, Grab the Special Discount - Save 25% - Ends In 00:00:00 Coupon code: SAVE25
Welcome to Pass4Success

- Free Preparation Discussions

Amazon Exam MLS-C01 Topic 3 Question 111 Discussion

Actual exam question for Amazon's MLS-C01 exam
Question #: 111
Topic #: 3
[All MLS-C01 Questions]

An ecommerce company wants to train a large image classification model with 10.000 classes. The company runs multiple model training iterations and needs to minimize operational overhead and cost. The company also needs to avoid loss of work and model retraining.

Which solution will meet these requirements?

Show Suggested Answer Hide Answer
Suggested Answer: D

Amazon SageMaker managed spot training allows for cost-effective training by utilizing Spot Instances, which are lower-cost EC2 instances that can be interrupted when demand is high. By enabling checkpointing in SageMaker, the company can save intermediate model states to Amazon S3, allowing training to resume from the last checkpoint if interrupted. This solution minimizes operational overhead by automating the checkpointing process and resuming work after interruptions, reducing the need for retraining from scratch.

This setup provides a reliable and cost-efficient approach to training large models with minimal operational overhead and risk of data loss.


Contribute your Thoughts:

Josefa
1 months ago
Hold up, are we sure we can't just train this model on a single beefy EC2 instance? I mean, how hard can 10,000 classes be, right? *laughs nervously*
upvoted 0 times
Salena
23 days ago
B) Use Amazon EC2 Spot Instances to run the training jobs. Use a Spot Instance interruption notice to save a snapshot of the model to Amazon S3 before an instance is terminated.
upvoted 0 times
...
Galen
26 days ago
A) Create the training jobs as AWS Batch jobs that use Amazon EC2 Spot Instances in a managed compute environment.
upvoted 0 times
...
...
Mari
2 months ago
I'm putting my money on C. Using Lambda to run the training jobs and saving the model weights to S3? That's some next-level cloud-native goodness right there.
upvoted 0 times
Victor
12 days ago
D) Use managed spot training in Amazon SageMaker. Launch the training jobs with checkpointing enabled.
upvoted 0 times
...
Thaddeus
24 days ago
C) Use AWS Lambda to run the training jobs. Save model weights to Amazon S3.
upvoted 0 times
...
Salena
25 days ago
B) Use Amazon EC2 Spot Instances to run the training jobs. Use a Spot Instance interruption notice to save a snapshot of the model to Amazon S3 before an instance is terminated.
upvoted 0 times
...
Buddy
1 months ago
A) Create the training jobs as AWS Batch jobs that use Amazon EC2 Spot Instances in a managed compute environment.
upvoted 0 times
...
...
Solange
2 months ago
That's true, SageMaker could also be a good option for minimizing operational overhead.
upvoted 0 times
...
Latonia
2 months ago
Hmm, I'm torn between B and D. Saving the model to S3 before instances get terminated is a nice safety net, but the SageMaker option seems more seamless. Decisions, decisions...
upvoted 0 times
Lauran
23 days ago
Agreed. It's a tough choice, but either way, the company needs to prioritize minimizing operational overhead and cost.
upvoted 0 times
...
Mitsue
24 days ago
True, both options have their benefits. It really depends on what the company values more - cost efficiency or seamless operation.
upvoted 0 times
...
Buck
28 days ago
But B also has its advantages. Saving a snapshot of the model to S3 before termination can prevent loss of work.
upvoted 0 times
...
Laura
1 months ago
I think D is the way to go. Managed spot training in SageMaker with checkpointing sounds efficient.
upvoted 0 times
...
...
Zack
2 months ago
Option D definitely seems like the way to go. Managed spot training in SageMaker with checkpointing? That's the perfect combo for this use case. Gotta love that AWS magic.
upvoted 0 times
Justine
1 months ago
I agree, using managed spot training in Amazon SageMaker with checkpointing enabled would definitely help minimize operational overhead and cost while avoiding loss of work and model retraining.
upvoted 0 times
...
Keneth
2 months ago
Option D definitely seems like the way to go. Managed spot training in SageMaker with checkpointing? That's the perfect combo for this use case. Gotta love that AWS magic.
upvoted 0 times
...
...
Denny
2 months ago
But what about option D with Amazon SageMaker? It has checkpointing enabled.
upvoted 0 times
...
Queen
2 months ago
I agree, using AWS Batch with Spot Instances can help minimize costs.
upvoted 0 times
...
Solange
3 months ago
I think option A sounds like a good choice.
upvoted 0 times
...

Save Cancel