stablediffusion
stablediffusion copied to clipboard
Stability-AI
What's the idea behind asymmetric padding during downsampling?
What's the idea behind this down-sampling with asymmetric padding : Why don't we just use a symmetric padding of 1, everything would fit perfectly.
class Downsample(nn.Module):
def __init__(self, in_channels, with_conv):
super().__init__()
self.with_conv = with_conv
if self.with_conv:
# no asymmetric padding in torch conv, must do it ourselves
self.conv = torch.nn.Conv2d(in_channels,
in_channels,
kernel_size=3,
stride=2,
padding=0)
def forward(self, x):
if self.with_conv:
pad = (0,1,0,1)
x = torch.nn.functional.pad(x, pad, mode="constant", value=0)
x = self.conv(x)
else:
x = torch.nn.functional.avg_pool2d(x, kernel_size=2, stride=2)
return x
Did you get an answer to this question? I have been wondering myself.
What's the idea behind this down-sampling with asymmetric padding : Why don't we just use a symmetric padding of 1, everything would fit perfectly.
class Downsample(nn.Module): def __init__(self, in_channels, with_conv): super().__init__() self.with_conv = with_conv if self.with_conv: # no asymmetric padding in torch conv, must do it ourselves self.conv = torch.nn.Conv2d(in_channels, in_channels, kernel_size=3, stride=2, padding=0) def forward(self, x): if self.with_conv: pad = (0,1,0,1) x = torch.nn.functional.pad(x, pad, mode="constant", value=0) x = self.conv(x) else: x = torch.nn.functional.avg_pool2d(x, kernel_size=2, stride=2) return x
I also wonder why it is.
