Learn how to quickly train an AI model at scale using Azure Machine Learning

AI and Machine learning are transforming science, industry and business with an always expanding range of applications. The pace of progress is relentless and with models becoming ever more complex and datasets ever larger, a single GPU or even a single machine with multiple GPUs is often not enough. Distributed training on large GPU clusters is becoming a more common requirement. For many organizations, owning such a cluster is not the best solution, so the cloud is a natural way to access large GPU clusters. In this tutorial we will show you how to quickly train a distributed model on your own GPU cluster using Azure Machine Learning.

It’s no secret that cloud computing can be complex, especially when directly managing infrastructure such as VMs and virtual networks. However, by using appropriate managed services the underlying infrastructure management is handled by the cloud platform. The Azure Machine Learning service allows the user to manage all aspects of the training (or inference) being performed, while automatically managing the underlying infrastructure.

By the end of this tutorial, you will understand how to create an AzureML workspace and configure datasets, software environments and compute resources. You will also learn how to create and submit training jobs using the AzureML Python SDK.