m = nn.ConvTranspose2d(1, 1, (2, 2), stride=(2, 2), padding=(1, 1), output_padding=(1,1))
output = m(input)
print(output)
tensor([[[[ 0.3960, 0.6212, 0.3960, 0.6212, 0.3960, 0.6212, 0.3960],
[-0.2565, 0.0108, -0.2565, 0.0108, -0.2565, 0.0108, -0.2565],
[ 0.3960, 0.6212, 0.3960, 0.6212, 0.3960, 0.6212, 0.3960],
[-0.2565, 0.0108, -0.2565, 0.0108, -0.2565, 0.0108, -0.2565],
[ 0.3960, 0.6212, 0.3960, 0.6212, 0.3960, 0.6212, 0.3960],
[-0.2565, 0.0108, -0.2565, 0.0108, -0.2565, 0.0108, -0.2565],
[ 0.3960, 0.6212, 0.3960, 0.6212, 0.3960, 0.6212, 0.3960]]]],
why output_padding not padding 0 to out?What number is pad?
How can I replace ConvTranspose2d（output_padding） with ConvTranspose2d(no output_padding)?I want to ConvTranspose2d with output_padding in MXNet ?