Difference in latent size between DCGAN and QGAN have on the training time

Is it feasible to build a quantum version of the DCGAN? A high latent size vector for noise is used in DCGAN. But in QGAN, the latent size is small. Will it change the time required for the training?

Hey @mass_of_15,

In principle, it should be doable! Because TorchLayer andKerasLayer` trick Torch and Keras/TF, respectively, into thinking a quantum circuit is a regular neural network layer, any classical neural network architecture can be made quantum to varying degrees.

Of course, the larger your network is, you can generally expect the training time to be longer.