ONNX Runtime is a cross-platform inference and training machine-learning
accelerator.
ONNX Runtime inference can enable faster customer experiences and lower costs,
supporting models from deep learning frameworks such as PyTorch and
TensorFlow/Keras as well as classical machine learning libraries such as
scikit-learn, LightGBM, XGBoost, etc. ONNX Runtime is compatible with different
hardware, drivers, and operating systems, and provides optimal performance by
leveraging hardware accelerators where applicable alongside graph optimizations
and transforms.
ONNX Runtime training can accelerate the model training time on multi-node
NVIDIA GPUs for transformer models with a one-line addition for existing PyTorch
training scripts.
NOTE: If this package has multiple flavors (see below), then use one of them instead of the name specified above. NOTE: This is a Python port. Instead of py312-onnxruntime listed in the above command, you can pick from the names under the Packages section.
misc/py-onnxruntime: Fix wheel missing the compiled pybind extension
setup.py's per-platform "libs" list decides which capi/ files get bundled
into the wheel as package data. It has dedicated branches for Linux
(manylinux and non-manylinux), Darwin, and Windows, but none for FreeBSD --
which silently fell through to the Windows-oriented default, whose *.dll/
*.pyd glob patterns never match anything here. The wheel built without
error, but ended up missing onnxruntime_pybind11_state.so entirely, along
with libonnxruntime_providers_shared.so and the versioned libonnxruntime.so;
the port previously worked around this with a manual post-install step that
copied the pybind extension in from the raw build tree afterward.
Add a FreeBSD branch to setup.py's libs list, mirroring the already-working
Darwin one (plain package-data entries, no ext_modules), so the wheel now
includes the same compiled artifacts on FreeBSD as it does on every other
(Only the first 15 lines of the commit message are shown above )
misc/py-onnxruntime: fix build on powerpc64le
/wrkdirs/usr/ports/misc/py-onnxruntime/work-py312/build/Release/_deps/eigen3-src/unsupported/Eigen/CXX11/../../../Eigen/src/Core/arch/AltiVec/PacketMath.h:119:35:
error: 'vec_lvsl' is deprecated: use assignment for unaligned little endian
loads/stores [-Werror,-Wdeprecated-declarations]
