pytorch_pfn_extras.onnx.grad#

pytorch_pfn_extras.onnx.grad(output, inputs, retain_graph=None, create_graph=False, only_inputs=True, allow_unused=False)#
Parameters:

  • output (Tensor) –

  • inputs (Tuple[Tensor, ...]) –

  • retain_graph (Optional[bool]) –

  • create_graph (bool) –

  • only_inputs (bool) –

  • allow_unused (bool) –

Return type:

Tuple[Optional[Tensor], …]