About the computation power to run the default setting of FAT

[momo1986]

Hello,

Thanks for your sharing and your outstanding contributions for adversarial learning.

I wonder how much computation power is needed to run the default settings, e.g., batch_size=128.

Could you list the corresponding GPU architectures which is able to run it? How many GPU memories are needed.

Thanks & Regards!
Momo

[GodXuxilie]

Thanks for your interest in our work.

12G GPU memories are needed to run the default settings in which batch size is 128 and the training model is Wide ResNet. If the training model is ResNet-18 and other settings are by default, 6G GPU memories are needed. In our paper, all the experiments are done on TITAN RTX (24G) graphic cards.

Let me know if you have any other questions.

Best wishes,
Xilie