8x for Super-FAN

[yinyiyu]

Hi Jzijin， I find that you hard code for x4 in your model/model.py since there are only two up layers, so I add one up layer and one residual layer for my x8 solution.Unfortunately, I encountered an GPU memory overflow problem, which occurred when the first epoch was trained up to about 90 iterations. I've tried to reduce the size of the batch size， but I still end up with a memory overflow. The I trained on the 2080ti which has 11G memory。 I don't know how it happed, can you help me !