Posted On: Oct 16, 2019
AWS Deep Learning (DL) Containers now support PyTorch. AWS DL Containers are Docker images pre-installed with deep learning frameworks to make it easy to setup and deploy custom machine learning environments. Docker images for training and inference with PyTorch are now available through HAQM Elastic Container Registry (HAQM ECR) free of charge—you pay only for the resources that you use. You can deploy AWS DL Containers on HAQM SageMaker, HAQM Elastic Kubernetes Service (HAQM EKS), self-managed Kubernetes on HAQM EC2, and HAQM Elastic Container Service (HAQM ECS).
AWS DL Containers are continually updated with the latest deep learning libraries and optimized for CPU and GPU resources on AWS. For more information, see AWS Deep Learning Containers. You can get started with AWS DL Containers for PyTorch here and review release notes and the bill of materials for Docker images here.