Differences with torch.isclose

torch.isclose

torch.isclose(input, other, rtol=1e-05, atol=1e-08, equal_nan=False) -> Tensor

For more information, see torch.isclose。

mindspore.ops.isclose(x1, x2, rtol=1e-05, atol=1e-08, equal_nan=False) -> Tensor

For more information, see mindspore.ops.isclose。

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

PyTorch: The dtype of parameter input and other can be bool , int or float .

MindSpore： The dtype of parameter x1 and x2 can be int32 , float32 or float16 .

Categories	Subcategories	PyTorch	MindSpore	Difference
Parameters	Parameter 1	input	x1	Different parameter names. The parameter `input` and `x1` are Tensors，while the dtype of `input` can be `bool` , `int` or `float` , the dtype of `x1` can be `int32` , `float32` or `float16` .
	Parameter 2	other	x2	Different parameter names. The parameter `other` and `x2` are Tensors，while the dtype of `other` can be `bool` , `int` or `float` , the dtype of `x2` can be `int32` , `float32` or `float16` .
	Parameter 3	rtol	rtol	-
	Parameter 4	atol	atol	-
	Parameter 5	equal_nan	equal_nan	-