Ask HN: Would Quantum GPUs be much better for Deep Learning than classical GPUs?

Please excuse any mislabeling of the terminology. My question is this. Would a quantum processing unit with a similar number of cores/quibits as a high end graphics card be much better at deep learning/ML tasks? It seems to me the ability to occupy super positions would allow a neural network to be much more elaborate than that which is allowed to exist in a classical computing system since in essence all a neural network does is make a ton of guesses then prunes the solutions that are not accurate representations of the data input. If quantum is not capable of doing this on its own or does not excel at it would it be possible to overlay quantum processor units and classical units in some kind of grid system like a checkerboard to leverage the strength of the other processing technology? Does anyone have any source info on this?