This post originally appeared on research.fb.com.
In September, we released an early version of the Open Neural Network Exchange format (ONNX) with a call to the community to join us and help create an open, flexible standard to enable deep learning frameworks and tools to interoperate. Today Facebook, AWS, and Microsoft are excited to announce that with the support of the community and new partners the first version of ONNX is now production-ready.
With ONNX, we are working to create an AI ecosystem that gives developers the freedom to innovate by providing the ability to combine tools and easy transfer models. We have been excited to see the immediate response the community and industry have given to this vision. Since launch, AMD, ARM, IBM, Intel, Huawei, NVIDIA, and Qualcomm have all announced support for ONNX, and many others in the community have contributed to the project.
Support for ONNX is available now in many top frameworks and runtimes including Caffe2, Microsoft’s Cognitive Toolkit, Apache MXNet, PyTorch and NVIDIA’s TensorRT. We also have community contributed converters for other projects such as TensorFlow. The current version of ONNX is design to work for most vision applications. For future versions, we are working together with ONNX partners and community to expand ONNX to represent scenarios beyond vision, including more dynamic models that occur in areas like language modeling. We hope others will get involved, contribute and help grow the ONNX ecosystem.