Which tool helps find training problems such as vanishing gradients?
Why C is the answer
SageMaker Debugger
Debugger inspects training jobs and flags common issues.
In detail
- Debugger finds training problems such as vanishing gradients.
- AWS closed SageMaker Debugger to new customers. Existing customers can keep using it.
- Use built-in rules for common issues.
Use case
An illustrative scenario, not a real incident.
A network stops learning, and Debugger flags vanishing gradients in the training job.
See it in different ways
mindmap
root((SageMaker AI))
Data
Data Wrangler
Ground Truth
Clarify
Train
JumpStart
Automatic model tuning
Debugger
Deploy
Pipelines
Model Registry
Operate
Model Monitorflowchart LR S0["Run with Debugger"] --> S1["Rules check tensors"] --> S2["Issue flagged"] --> S3["Fix"]
Key point: Debugger finds training problems.
Why the other options are wrong
- A. SageMaker Clarify
- Clarify checks bias and explains models.
- B. SageMaker Ground Truth
- Ground Truth labels data.
- D. SageMaker Model Monitor
- Model Monitor watches production.