AIM216-R1 – Amazon SageMaker Debugger – Insights into ML model (re:Invent 2019) – Key Takeaways

The Key

  • Debugging the training process can be difficult, SageMaker Debugger can solve the problem with minimal efforts

The Takeaways

  • Traditional way of testing layers using logs, complex and error-prone
  • SageMaker Debugger
    • Rules = assertions
    • Built-in rules are free
    • No code changes needed
  • Change Healthcare case study
    • AI-driven claim form processing
    • Use SageMaker Debugger to analyze layers and tensors, use rules to early stop (break the training if model outputs blank images)
    • 55% reduction in human-processed claims

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s