Amazon SageMaker now supports TensorFlow 1.5, Apache MXNet 1.0, and CUDA 9 for P3 Instance Optimization
You can now use TensorFlow 1.5 and Apache MXNet 1.0 with the pre-built deep learning framework containers that are provided with Amazon SageMaker. Amazon SageMaker also now supports CUDA 9 drivers, for faster performance on Amazon SageMaker ml.p3 instances. Both updates to TensorFlow and MXNet provide improved performance features, among other details that can be found on the respective GitHub repositories for TensorFlow 1.5 and MXNet 1.0.0.
Support for CUDA 9 and the latest TensorFlow and MXNet updates in Amazon SageMaker is available today in the US East (N. Virginia & Ohio), EU (Ireland), and U.S. West (Oregon) AWS regions. To learn more about how to use these newly released containers in Amazon SageMaker, visit the sample notebooks on GitHub.