Hi, I'm currently trying to train the LSUN model(256x256). When using a 8xA100(80G) machine, however, I can only set the global the batch size of 64 instead of 256 mentioned in the paper using the default model architecture. I'm wondering if I'm missing something here. Thank you in advance for the help!
Hi, I'm currently trying to train the LSUN model(256x256). When using a 8xA100(80G) machine, however, I can only set the global the batch size of 64 instead of 256 mentioned in the paper using the default model architecture. I'm wondering if I'm missing something here. Thank you in advance for the help!