Padding 'same' exists for Conv1d but not for ConvTranspose1d

vision
1

Hi,
In Conv1d, the padding parameter can take the value of same. This doesn’t work for convTranpose1d. What is the best way to achieve this:

conv1 = .nn.Conv1d(3, 16, kernel_size=25, padding='same')
x = torch.randn(4, 3, 15)
out = conv1(x)
>>> out.shape
torch.Size([4, 16, 15])
conv1dtranspose = nn.ConvTranspose1d(16, 3, kernel_size=25)
v = conv1dtranspose(out)
>>> v.shape
torch.Size([4, 3, 39])

What’s the best way to get torch.Size([4, 3, 15]) from conv1dtranspose?
Thank you