Hi, Thanks for sharing this nice work! I have some questions about training details of VQGAN on ImageNet. - How many epochs should I train? - How many GPUs are used? (1?) What should the batch size be set to? (12?) - How many days it cost? Looking forward to reply.