neuralop.layers.skip_connections.SoftGating

class neuralop.layers.skip_connections.SoftGating(in_features, out_features=None, n_dim=2, bias=False)[source]

Applies soft-gating by weighting the channels of the given input

Given an input x of size (batch-size, channels, height, width), this returns x * w ` where w is of shape `(1, channels, 1, 1)

Parameters:

in_featuresint

out_featuresNone

this is provided for API compatibility with nn.Linear only

n_dimint, default is 2

Dimensionality of the input (excluding batch-size and channels). n_dim=2 corresponds to having Module2D.

biasbool, default is False

Methods

forward(x)

Applies soft-gating to a batch of activations

forward(x)[source]

Applies soft-gating to a batch of activations