All terms
Frameworks & Tools
ONNX
An open standard for representing trained models so they run across tools and hardware.
Definition
ONNX is an open standard for representing trained machine learning models in a framework-neutral format. A model built in one tool — such as PyTorch — can be exported to ONNX and then run in another framework or on an optimized runtime. It is widely used to deploy models across diverse hardware without rewriting them.