Step 1: Modify Your Own Training Script Using SageMaker's Distributed Model Parallel Library - Amazon SageMaker

Step 1: Modify Your Own Training Script Using SageMaker's Distributed Model Parallel Library

Use this section to learn how to customize your training script to use the core features of the Amazon SageMaker distributed model parallel library. To use the library-specific API functions and parameters, we recommend you use this documentation alongside the SageMaker model parallel library APIs in the SageMaker Python SDK documentation.

The training script examples provided in these sections are simplified and designed to highlight the required changes you must make to use the library. For end-to-end, runnable notebook examples that demonstrate how to use a TensorFlow or PyTorch training script with the SageMaker distributed model parallel library, see Amazon SageMaker Distributed Training Notebook Examples.