Note

  This page describes an internal API which is not intended to be used outside of the PyTorch codebase and can be modified or removed without notice.

ONNX Backend for TorchDynamo

For a quick overview of torch.compiler, see torch.compiler.

Warning

The ONNX backend for torch.compile is a rapidly evolving beta technology.

torch.onnx.is_onnxrt_backend_supported()

Returns True if ONNX Runtime dependencies are installed and usable to support TorchDynamo backend integration; False otherwise.

Example:

# xdoctest: +REQUIRES(env:TORCH_DOCTEST_ONNX)
>>> import torch
>>> if torch.onnx.is_onnxrt_backend_supported():
...     @torch.compile(backend="onnxrt")
...     def f(x):
...             return x * x
...     print(f(torch.randn(10)))
... else:
...     print("pip install onnx onnxscript onnxruntime")
...

Return type

bool