python3 main.py --phase train --gpu_num 2 --global_batch_size 8
- --gpu_num: the number of gpu you use
- --global_batch_size: overall batch size (which means each batch size per gpu is global_batch_size / gpu_num)
- If you change other parameters, check config.json.
python3 main.py --phase test --gpu_num 2 --global_batch_size 8
- Options are the same as training.
Reference the base implementation: https://github.com/xhujoy/CycleGAN-tensorflow.git