sdpython/onnx-extended
New operators for the ReferenceEvaluator, new kernels for onnxruntime, CPU, CUDA
This project helps machine learning engineers and data scientists accelerate their ONNX models by replacing standard operator implementations with highly optimized C++ versions. You provide an ONNX model, and it outputs the same model running significantly faster, especially on small graphs and tensors. It's designed for those deploying or working with ONNX models who need maximum inference speed.
Available on PyPI.
Use this if you need to speed up inference for your ONNX models by leveraging faster, custom C++ implementations or new, extended operators.
Not ideal if you are developing or training models in frameworks like PyTorch or TensorFlow and don't yet have an ONNX model.
Stars
35
Forks
6
Language
Python
License
MIT
Category
Last pushed
Feb 13, 2026
Commits (30d)
0
Dependencies
3
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/sdpython/onnx-extended"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Related frameworks
microsoft/onnxruntime
ONNX Runtime: cross-platform, high performance ML inferencing and training accelerator
onnx/onnx
Open standard for machine learning interoperability
PINTO0309/onnx2tf
Self-Created Tools to convert ONNX files (NCHW) to TensorFlow/TFLite/Keras format (NHWC). The...
NVIDIA/TensorRT
NVIDIA® TensorRT™ is an SDK for high-performance deep learning inference on NVIDIA GPUs. This...
onnx/onnxmltools
ONNXMLTools enables conversion of models to ONNX