AWS Machine Learning Blog

Announcing the Amazon SageMaker MXNet 1.2 container

The Amazon SageMaker pre-built MXNet container now uses the latest release of Apache MXNet 1.2.  Amazon SageMaker is a fully-managed platform that enables developers and data scientists to quickly and easily build, train, and deploy machine learning models at any scale.  And the pre-built MXNet container makes it easy to write your deep learning scripts naturally but still take advantage of distributed, managed training and real-time production hosting in Amazon SageMaker.

MXNet 1.2 brings ease of use, faster performance, and enhanced interoperability. It also adds the new Intel MKL-DNN integration that accelerates neural network operators such as convolution, deconvolution, and pooling on compute-optimized C5 instances. The enhanced FP16 support accelerates mixed precision training on Tensor Cores of NVIDIA Volta V100 GPUs powering Amazon EC2 P3 instances. Finally, MXNet 1.2 comes with a new Open Neural Network Exchange Format (ONNX) module for easily importing ONNX models into the MXNet symbolic interface.

ONNX enables interoperability between deep learning frameworks, including PyTorch, Caffe2, Microsoft Cognitive Toolkit (CNTK), Chainer, and MXNet.  And the ONNX model zoo makes it easy to download a variety of pre-trained, high-quality models to generate predictions for tasks like image classification, object detection, semantic segmentation, super resolution, face emotion recognition, and more.  A previous AWS machine learning blog post showed how easy it is to import and work with these models.  Combining this with Amazon SageMaker hosting, makes it easy for you to integrate machine learning predictions into your service.  To learn more about deploying an ONNX model with the Amazon SageMaker MXNet container see this example notebook.

The Amazon SageMaker MXNet 1.2 container is now available in the following AWS Regions: US East (N. Virginia), US East (Ohio), US West (Oregon), Europe (Ireland), Europe (Frankfurt), Asia Pacific (Tokyo), Asia Pacific (Seoul), and Asia Pacific (Sydney). See the Amazon SageMaker documentation and the sagemaker-mxnet-container repository for more details.

About the Author

David Arpin is AWS’s AI Platforms Selection Leader and has a background in managing Data Science teams and Product Management.