In Table A10 of the paper on Arxiv. I see experiments on Swin-T architecture. Could you clarify which dataset is used to conduct conduct those experiments? Is it CIFAR-10? If yes, then what modification did you do to the architecture hyperparameters to train on 32x32 images? Thanks in advance.
In Table A10 of the paper on Arxiv. I see experiments on Swin-T architecture. Could you clarify which dataset is used to conduct conduct those experiments? Is it CIFAR-10? If yes, then what modification did you do to the architecture hyperparameters to train on 32x32 images? Thanks in advance.