ONNX

ONNX (Open Neural Network Exchange) is an open standard that defines a common set of operators and a file format for machine learning models. A model trained in one framework can be exported to ONNX and run on many different runtimes, compilers and hardware accelerators.

Developers use ONNX to separate training from deployment: they keep working in their preferred framework and pick the inference engine that suits their hardware. Compatible runtimes and libraries apply hardware-specific optimizations, and the site lists the supported frameworks and accelerators.

ONNX is a community project with open governance and is a graduate project of LF AI and Data. Work happens through special interest groups and working groups, with discussion on Slack, a contribution guide and regular community meetups.