torch.isinf

torch.isinf(input) → Tensor

Tests if each element of input is infinite (positive or negative infinity) or not.

Note

Complex values are infinite when their real or imaginary part is infinite.

Parameters:

input (Tensor) – the input tensor.

Returns:

A boolean tensor that is True where input is infinite and False elsewhere

Example:

>>> torch.isinf(torch.tensor([1, float('inf'), 2, float('-inf'), float('nan')]))
tensor([False,  True,  False,  True,  False])