onnx/tutorials

This repository is a collection of tutorials for working with ONNX, which stands for Open Neural Network Exchange. ONNX is an open standard format for saving and sharing machine learning models. The core idea is that a model trained in one framework, such as PyTorch or TensorFlow, can be exported to the ONNX format and then loaded and run in a completely different framework or runtime. This makes it easier to train a model in one environment and deploy it in another, or to switch tools without retraining from scratch. The tutorials are organized around two main tasks. The first is converting models from their native format into ONNX. The repository includes example notebooks for converting models from PyTorch, TensorFlow, Keras, Scikit-Learn, MXNet, Caffe2, MATLAB, and several other frameworks. Each entry points to an installation guide and a working code example. The second task is running, or "scoring," an ONNX model after it has been converted. Supported runtimes for this include ONNX Runtime, Caffe2, CoreML on Apple devices, TensorRT for Nvidia hardware, Windows ML, and others. The tutorials themselves are primarily Jupyter Notebooks, which are interactive documents that mix code, explanations, and output in a format that can be opened and run step by step. This makes them practical for someone who wants to follow along and try the conversions themselves. Beyond conversion, the repository also points to the ONNX Model Zoo, which is a separate collection of pre-trained ONNX models ready to use for common tasks. A section covers end-to-end examples that go from training to deployment, including tasks like image classification, object detection, and language processing. The repository is maintained by the ONNX organization on GitHub and reflects a broad ecosystem of tools that have adopted ONNX as a common interchange format for machine learning models.