Amazon SageMaker Now Supports PyTorch and TensorFlow 1.8
Amazon SageMaker now comes pre-configured to run PyTorch, adding to the existing integrated TensorFlow, Apache MXNet and Chainer deep learning frameworks that are currently available. Additionally, the pre-configured TensorFlow containers in Amazon SageMaker now support versions 1.7 and 1.8.
Amazon SageMaker now comes pre-configured to run PyTorch, adding to the existing integrated TensorFlow, Apache MXNet and Chainer deep learning frameworks that are currently available. Additionally, the pre-configured TensorFlow containers in Amazon SageMaker now support versions 1.7 and 1.8.
Using PyTorch in Amazon SageMaker is as easy as using the other pre-built deep learning frameworks. Since PyTorch is deeply integrated with Python, it allows you to use typical Python control flows in your networks. PyTorch also supports dynamic computation graphs, which allow for more flexible use of memory and better support for recursive computations. You can read details about PyTorch in the blog post here.
Pre-built containers for PyTorch and TensorFlow 1.7 and 1.8 for Amazon Sagemaker are now available in the US East (N. Virginia), US East (Ohio), US West (Oregon), Europe (Ireland), and Asia Pacific (Tokyo) AWS regions. Visit the documentation for more information on PyTorch.