Adapt single node PyTorch to distributed deep learning
Learn how to perform distributed training of machine learning models using PyTorch.
This notebook follows the recommended development workflow. It first shows how to train the model on a single node, and then how to adapt the code using HorovodRunner for distributed training.
HorovodRunner PyTorch MNIST example notebook
Feedback
https://aka.ms/ContentUserFeedback.
Coming soon: Throughout 2024 we will be phasing out GitHub Issues as the feedback mechanism for content and replacing it with a new feedback system. For more information see:Submit and view feedback for