biclab / asa-snn Goto Github PK

Thank you for your excellent work!

I'm somewhat confused about the gradient calculation of the mask operation. It seems that input of the get_im_subim_channels function includes channels_im and map_add, while the output is computed by using top-k for map_add and generates binary mask. So, how does the gradient backward to map_add? And How are the weights of self.fc updated?

I cannot find 'TAda' in models.module. Is that a missing module? Will you add it into the codes pls?~~~

请问gait-day应该怎么转成.npy呢

您方便公布一下，各种数据集的预处理代码嘛？

非常感谢贵团队为我学习的领域作出的巨大贡献，我在gait数据集上复现时，发现下载的数据集是txt文件，而代码要求是npy文件，请问应该怎样转换呢，可以公布你们处理的代码吗？