Dropout3d

Dropout3d(p=0.5, inplace=False)

Zero whole channels of a channels-last 3-d feature map,
``(N, D, H, W, C)`` or ``(D, H, W, C)``, with probability ``p`` and
scale the rest by ``1 / (1 - p)`` while training; the identity in
evaluation mode.

forward

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

forward(input) -> Tensor