AlphaDropout

AlphaDropout(p=0.5, inplace=False)

The self-normalizing dropout for SELU networks: while training, a
dropped element takes the activation's saturation value and the result
is rescaled and shifted so the mean and variance stay; the identity in
evaluation mode.

forward

forward(self, input: 'Tensor') -> 'Tensor'

forward(input) -> Tensor