Errror when using train_asmlp_b.sh

[HantingChen]

When using train_asmlp_b.sh, the above error occurs:

File "/home/ma-user/work/AS-MLP-main/models/as_mlp.py", line 165, in forward
x = shortcut + self.drop_path(x)
File "/home/ma-user/anaconda3/envs/Pytorch-1.4.0/lib/python3.6/site-packages/apex/amp/wrap.py", line 58, in wrapper
return orig_fn(*new_args, **kwargs)
RuntimeError: The size of tensor a (56) must match the size of tensor b (60) at non-singleton dimension 3

[wp2333]

I get this error too. did you solve it?

[HantingChen]

I get this error too. did you solve it?

I am trying to fix it. It seems that the AxialShift block changes the size (

AS-MLP/models/as_mlp.py

Line 162 in f9b4000

x = self.axial_shift(x) # nW*B, window_size*window_size, C

).

They pad the input (

AS-MLP/models/as_mlp.py

Line 69 in f9b4000

x = F.pad(x, (self.pad, self.pad, self.pad, self.pad) , "constant", 0)

) but without shrinking it.

[niujinshuchong]

@HantingChen @wp2333
Thanks for pointing this. Yes, we should comment this line https://github.com/svip-lab/AS-MLP/blob/main/models/as_mlp.py#L70.

I updated the code and it should solve this error.