onnx/onnx

Health v2 · maintenance only

Adoption not part of the score

21366 stars · 4011 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

ONNX (Open Neural Network Exchange) is an open standard and Python library for representing machine learning and deep learning models as an extensible computation graph with defined operators and data types. It enables model interoperability between frameworks like PyTorch, TensorFlow, and scikit-learn and across diverse inference runtimes and hardware.

Use cases

• convert a pytorch model to run in a different inference engine
• export tensorflow models to a portable format
• run machine learning models on edge hardware
• share trained models between frameworks
• deploy models to production with hardware-optimized runtimes
• inspect and validate neural network computation graphs

When to choose

• you need to move models between training frameworks and inference runtimes
• you want hardware-accelerated deployment across vendors
• you need a stable, widely supported open model format

When to avoid

• you only train and infer within a single framework with no portability needs
• you need training-time features, since ONNX focuses on inferencing
• you need framework-specific optimizations not expressible in the ONNX operator set

Facets

library · maturity stable

machine-learning deep-learning serialization sdk machine-learning deep-learning artificial-intelligence developer-tools python cross-platform onnx model-format interoperability neural-networks inference computation-graph pytorch tensorflow

Member repositories

For agents

markdown · JSON · MCP: product_card(name="onnx/onnx")