Why using 4096 input vector size?

[artucalvo]

Great work, but I am trying to understand why are you using 4096 as input vector size, while in the DCGAN paper it uses 100.

def noise(n):
    return np.random.uniform(-1.0, 1.0, size = [n, 4096])

# And in the generator
 self.G.add(Reshape(target_shape = [1, 1, 4096], input_shape = [4096]))

I tried to change those values to 100, but it gives error stating that reshape expects dimension 100.

Any thoughts?

[rohanbanerjee]

It is because the desired size is 256*256 and therefore we need more dimension, which is more and different from what is mentioned in the dcgan paper.