However I meet with difficulty that the Conv2DTranspose in tf can not be translated to pytorch,because the ConvTranspose2d in pytorch don't have argument padding "same".
Can I slice the tensor to correct shape?
Is there any effect on the result if i do this?
However I meet with difficulty that the Conv2DTranspose in tf can not be translated to pytorch,because the ConvTranspose2d in pytorch don't have argument padding "same". Can I slice the tensor to correct shape? Is there any effect on the result if i do this?