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])