MLA-C01 Question 373
Select 2You are a machine learning engineer working on a model training pipeline in Amazon SageMaker. During training, you notice that the training job fails due to insufficient compute resources. You need to modify your setup to ensure the training job runs successfully. What steps should you take to address this issue?
- A
Increase the instance type to one with more compute capacity in the SageMaker training job configuration.
- B
Enable Amazon SageMaker Debugger to analyze the resource utilization of the training job.
- C
Switch to a managed Spot Training job to automatically handle compute resource scaling.
- D
Monitor the CloudWatch metrics for the training job to identify compute bottlenecks.
- E
Add an Auto Scaling policy to the training job to dynamically provision compute resources.
Show answer and explanation
Correct answers: A, D
Explanation
Ensuring sufficient compute resources is critical for successful training in Amazon SageMaker. Increasing the instance type provides the necessary compute capacity, while monitoring CloudWatch metrics allows you to identify and confirm bottlenecks. Other options, such as enabling SageMaker Debugger or Spot Training, may provide additional insights or cost savings but do not directly resolve the issue of insufficient compute capacity.
- A. Correct.
Correct. Increasing the instance type to one with more compute capacity can directly address the issue of insufficient resources.
- B. Incorrect.
While enabling SageMaker Debugger can provide insights into resource utilization, it does not directly resolve the issue of insufficient compute resources.
- C. Incorrect.
Using managed Spot Training can reduce costs but does not address the problem of insufficient compute capacity for the training job.
- D. Correct.
Correct. Monitoring CloudWatch metrics can help you identify compute bottlenecks and validate the resource insufficiency issue.
- E. Incorrect.
Amazon SageMaker training jobs do not support Auto Scaling policies for dynamically provisioning compute resources during training.