onnx
Open standard for machine learning interoperability
What is onnx?
Open standard for machine learning interoperability
Overview of onnx
What it is
ONNX (Open Neural Network Exchange) is an open ecosystem that provides an open source format for AI models, covering both deep learning and traditional machine learning. It defines an extensible computation graph model, built-in operators, and standard data types, with a current focus on inferencing (scoring) capabilities. ONNX is widely supported across many frameworks, tools, and hardware, enabling interoperability between different tools and streamlining the path from research to production.
Key features
- Open source format for AI models (deep learning and traditional ML)
- Extensible computation graph model with built-in operators and standard data types
- Focus on inferencing (scoring) capabilities
- Widely supported across frameworks, tools, and hardware
- Enables interoperability between different frameworks
- Programming utilities: shape and type inference, graph optimization, opset version conversion
- Python API and community-driven open governance
Use cases
- Exchanging AI models between different frameworks (e.g., PyTorch, TensorFlow, MXNet)
- Streamlining the transition from research to production by using a common model representation
- Building AI pipelines that combine tools from different vendors or open-source projects
- Deploying models on diverse hardware platforms that support ONNX
Getting started
Install ONNX via pip:
pip install onnx
For optional reference implementation dependencies, use:
pip install onnx[reference]
Detailed install instructions, including common build options and troubleshooting, are available in the repository README.
License and pricing
ONNX is open source under the Apache License 2.0. It is free to use and modify.
Based on the project's public README and repository metadata.