Apache MXNet on AWS
Developer Guide

Set Up a Stack for Distributed Deep Learning Using Apache MXNet

When working with large amounts of data, running MXNet on a cluster of EC2 instances allows you to scale horizontally—to add as many EC2 instances as you need to provide the required compute power and memory. This topic explains how to set up a cluster of EC2 instances for distributed deep learning with MXNet on AWS. In addition to setting up the cluster using an AWS CloudFormation template and installing MXNet and its dependencies, you learn how to run distributed training on an image classification model using the CIFAR-10 dataset. For more information, see the CIFAR-10 dataset.