Differences with torch.log

torch.log

torch.log(input, *, out=None) -> Tensor

For more information, see torch.log.

mindspore.ops.log

mindspore.ops.log(input) -> Tensor

For more information, see mindspore.ops.log.

Differences

MindSpore API function is consistent with that of PyTorch, with differences in the data types supported by the parameters.

PyTorch: The dtype of the parameter input can be int or float.

MindSpore: The dtype of the parameter input can be float16, float32 or float64 on CPU platform, while the dtype of the parameter input can be float16 or float32 on Ascend platform.

Categories	Subcategories	PyTorch	MindSpore	Difference
Parameters	Parameter 1	input	input	Both are Tensor, and the dtype of the parameter `input` in torch.log can be `int` or `float`. The dtype of the parameter `input` in mindspore.ops.log can be `float16`, `float32`, `float64` on CPU platform, while `float16` or `float32` on Ascend platform.
	Parameter 2	out	-	For detailed, refer to General Difference Parameter Table.