Pytorch does not pad the layers

Question

While making a network I used deconvolution layer :

def deconv3d(cin,cout,k=4,s=2,pad=-1):
    pad = (k - 1) // 2 if pad < 0 else pad
    return nn.Sequent