Importing and Exporting Neural Networks with ONNX

ONNX is an open format to represent deep learning models. It allows easy conversion of models from one framework to another. In this talk, we will provide an overview of the current state of the ONNX importer and exporter featured in the Wolfram Language and show how to use their options and properties. With the help of a few models (that are currently in the Neural Net Repository or will be added soon), we will show how to import and export models from and to the popular deep learning frameworks (TensorFlow/PyTorch/Core ML). Finally, we will discuss what new features can be expected in the upcoming releases.