torchadf.nn.functional.avg_pool3d

torchadf.nn.functional.avg_pool3d(in_mean, in_var, kernel_size, stride=None, padding=0, ceil_mode=False, count_include_pad=True, divisor_override=None, mode='diag')

Applies an average pooling function for 3D inputs.

Assumed Density Filtering (ADF) version of torch.nn.functional.avg_pool3d.

Parameters:

in_meantorch.Tensor

Input mean tensor.

in_vartorch.Tensor

Input (co-)variance tensor.

kernel_sizeint or tuple of int

Size of the pooling window.

strideint or tuple of int, optional

Stride of the pooling window (Default = kernel_size).

paddingint or tuple of int, optional

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

ceil_modebool, optional

Use ceil instead of floor to compute output shape (Default False).

count_include_padbool, optional

Include zero-padding in average calculations (Default True).

divisor_overrideoptional,

Will be used as divisor if specified, otherwise kernel_size will be used (Default None).

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

Covariance propagation mode (Default “diag”).

Returns:

out_meantorch.Tensor

The transformed mean tensor.

out_vartorch.Tensor

The transformed (co-)variance tensor.