According to Microsoft, the ONXX runtime interface is going to provide a boost for the ONXX Runtime program. It is used to accelerate the training of machine learning. It also serves as a standard for ML interoperability.

The runtime will also let faster consumer experiences to pertain lower costs. It acts as a supporting system for models with deep learning frameworks like PyTorch, TensorFlow/Keras scikit-learn, LightGBM, XGBoost, and much more. The JavaScript-based ORT-Web for Runtime Process is a replacement over Microsoft's ONNX.js Javascript library for better UX and performance.

Other features of the new tool include improved experience for developers linking the backend and the frontend performance. It gives great model coverage to the developers running the machine learning process on browsers. In short, it enhances the server-client communication without the need for any further installations. All this is a part of the ecosystem of Open Neural Network Exchange (ONXX).

Refer to MindStick news for more stuff on Microsoft.