[trasposed convolution 에 대해 다뤘던 글]
torch.nn.ConvTranspose2d(in_channels, out_channels, kernel_size, stride=1, padding=0, output_padding=0, groups=1, bias=True, dilation=1, padding_mode='zeros', device=None, dtype=None)
여러 input 평면으로 이뤄진 input image에 2d transposed convolution 을 적용한다.
입력에 대한 Conv2d의 gradient로 볼 수 있다.
보통 Deep Convolution GAN을 구현할 때 주로 사용된다.
•
stride controls the stride for the cross-correlation.
•
padding controls the amount of implicit zero padding on both sides for dilation * (kernel_size - 1) - padding number of points.
•
output_padding controls the additional size added to one side of the output shape.
•
dilation controls the spacing between the kernel points; also known as the à trous algorithm. It is harder to describe, but the link here has a nice visualization of what dilation does.
•
groups controls the connections between inputs and outputs.
•
in_channels and out_channels must both be divisible by groups.
For example,
◦ At groups=1, all inputs are convolved to all outputs.
◦ At groups=2, the operation becomes equivalent to having two conv layers side by side, each seeing half the input channels and producing half the output channels, and both subsequently concatenated.
◦ At groups= in_channels, each input channel is convolved with its own set of filters (of size ).
The parameters kernel_size, stride, padding, output_padding can either be:
a single int – height와 width dimension에서 같은 값으로 사용
a tuple of two ints – 첫번째 값이 height dimension에서, 두번째 값이 width dimension에서 사용
parameters
•
•
•
groups (int, optional) – Number of blocked connections from input channels to output channels. Default: 1
•
shape
input
output
