c-pupil

I want to used the pre-trained encoder as my model encoder, and build use the code ` self.pretrained1 = torch.hub.load('torchhub/facebookresearch_dinov2_main', 'dinov2_{:}14'.format(encoder), source='local', pretrained=False`)” and feed the last four layers of...

Difference in memory usage of mlp and kan

1

Hello： if input tensor size is [64,28x28],hidden layers is [256,256,256,256],The memory usage of mlp and kan is similar,382M and 500M respectively.The results are consistent with the experimental [results:](https://github.com/Jerry-Master/KAN-benchmarking) However，if the...