Biggan-Deep Check

[gcervantes8]

I'm getting good results with Biggan-Deep model using the N64 game dataset but not with ImageNet. I'm not sure if this is normal because of my low batch size or if there is something wrong with the code.

Training an ImageNet model with a low batch size from: https://github.com/ajbrock/BigGAN-PyTorch might help me understand that.