📚 Documentation

Hey, there seems to be a simple mistake in the equation for the output of all the transposed convolution layers (e.g https://pytorch.org/docs/master/nn.html#convtranspose2d). The equations do not take into account dilation.

I believe the correct formula should be (in the 1d case)

(L_in - 1) * s - 2 * p + (k - 1)*d + o_p + 1

where L_in, s, p, k, d, o_p are input length, stride, padding, kernel size, dilation and output padding, respectively.