FeatureAlphaDropout

FeatureAlphaDropout(p=0.5, inplace=False)

Channel-wise alpha dropout over a channels-last input: whole channels
take the saturation value with probability ``p`` while training, with
the same rescale and shift as :class:`AlphaDropout`; the identity in
evaluation mode.

forward

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

forward(input) -> Tensor