MLS-C01 · Question #233
A machine learning (ML) specialist has prepared and used a custom container image with Amazon SageMaker to train an image classification model. The ML specialist is performing hyperparameter…
The correct answer is C. Use SageMaker Experiments to run and manage multiple training jobs and tune the model of the. SageMaker Experiments is designed exactly for this use case: it lets you organize, track, compare, and analyze multiple training jobs and HPO runs from within SageMaker Studio notebooks. You can create an experiment that groups HPO jobs from both the custom container and the…
Question
A machine learning (ML) specialist has prepared and used a custom container image with Amazon SageMaker to train an image classification model. The ML specialist is performing hyperparameter optimization (HPO) with this custom container image to produce a higher quality image classifier. The ML specialist needs to determine whether HPO with the SageMaker built-in image classification algorithm will produce a better model than the model produced by HPO with the custom container image. All ML experiments and HPO jobs must be invoked from scripts inside SageMaker Studio notebooks. How can the ML specialist meet these requirements in the LEAST amount of time?
Options
- APrepare a custom HPO script that runs multiple training jobs in SageMaker Studio in local mode
- BUse SageMaker Autopilot to tune the model of the custom container image. Use the automatic
- CUse SageMaker Experiments to run and manage multiple training jobs and tune the model of the
- DUse the automatic model tuning capability of SageMaker to tune the models of the custom
How the community answered
(21 responses)- A5% (1)
- B10% (2)
- C62% (13)
- D24% (5)
Explanation
SageMaker Experiments is designed exactly for this use case: it lets you organize, track, compare, and analyze multiple training jobs and HPO runs from within SageMaker Studio notebooks. You can create an experiment that groups HPO jobs from both the custom container and the built-in algorithm, then compare their metrics side by side with minimal overhead. Option A (local mode HPO script) would require writing substantial custom orchestration and lacks centralized tracking. Option B (Autopilot) automates algorithm selection and training but does not support comparing a custom container against a built-in algorithm - it manages its own models. Option D (automatic model tuning alone) can run HPO but lacks the experiment tracking and comparison capability needed to evaluate both approaches together.
Topics
Community Discussion
No community discussion yet for this question.