The Open Neural Network Exchange (ONNX) is a community project originally launched in September 2017 to increase interoperability between deep learning tools. ONNX is a standard for representing deep learning models that enables these models to be transferred between frameworks. It is the first step toward an open ecosystem where AI developers can easily move between state-of-the-art tools and choose the combination that works best for them.
Today we’re excited to announce that Amazon Web Services (AWS) is contributing ONNX support for Apache MXNet and AWS is joining the ONNX effort. By joining ONNX, AWS is contributing to this innovative open ecosystem for interchangeable AI models. Together, we’re giving developers the power to build and create rich experiences – without having to worry about interoperability across frameworks.
As the initial part of this collaboration, AWS will help build out the ONNX format for the broader deep learning community. We are excited to work together to make ONNX more accessible and robust for developers everywhere.
First Up: MXNet Support
AWS’s first initiative is MXNET support. ONNX-MXNet is an open source Python package designed to import ONNX deep learning models into Apache MXNet. As a fully featured deep learning framework, MXNet provides APIs across languages like Python, Scala, and R.
With MXNet support, developers will have the opportunity to build and train models with other frameworks, like PyTorch, Microsoft Cognitive Toolkit, or Caffe2. From there, they can import the models into MXNet to run them for inference, leveraging its optimized, scaleable engine.
Together, the ONNX partners and community will continue to develop the ONNX format and ecosystem. We will be adding more interoperability, expanding the ONNX MXNet functionality, and bringing ONNX into MXNet core APIs. Our mission is to give developers the opportunity work freely work across all frameworks. Visit the ONNX website to find out more.