torchadf.nn.modules.conv.Conv2d

class torchadf.nn.modules.conv.Conv2d(in_channels, out_channels, kernel_size, stride=1, padding=0, dilation=1, groups=1, bias=True, padding_mode='zeros', device=None, dtype=None, mode='diag')

A 2D convolution layer.

Assumed Density Filtering (ADF) version of torch.nn.Conv2d.

Parameters:

in_channelsint

Number of input channels.

out_channelsint

Number of output channels.

kernel_sizeint or tuple of int

Size of the convolution kernel.

strideint or tuple of int, optional

Stride of the convolution kernel (Default 1).

paddingint or tuple of int, optional

Implicit padding on both sides of input (Default 0).

dilationint or tuple of int, optional

Spacing between kernel elements (Default 1).

groupsint, optional

Split convolution into a number independent groups (Default 1). Number of input channels has to be divisible by the number of groups.

biasbool, optional

Add a learnable convolution bias (Default True).

padding_mode{“zeros”, “reflect”, “replicate”, “circular”}, optional

Padding mode (Default “zeros”).

mode{“diag”, “diagonal”, “lowrank”, “half”, “full”}, optional

Covariance propagation mode (Default “diag”).