Abstract: PyTorch is a popular open source machine learning framework. AzureML has launched a new container for simplifying and accelerating deep learning jobs with PyTorch. This container brings together PyTorch and Microsoft technologies for accelerating large scale distributed training and is used by many teams at Microsoft to develop the models used in Office, Bing, Dynamics, etc. In this talk, we discuss the contents of the PyTorch on Azure container and share the results seen by early adopters. In addition, we will show demos of how you can get started with the container and how to use the Microsoft acceleration technologies it comes with.
Bio: Jambay Kinley is a Software Engineer in Azure AI Platform organization at Microsoft. He is a member of a team that works on various Pytorch related initiatives for the AI Frameworks team. Prior to joining Microsoft, he was involved in Machine Learning research in a Natural Language Processing lab at Cornell Tech.