Is there an issue with the code for calculating the loss function

[gczzcc]

I noticed that there might be an issue with the code for calculating the final pose loss function in the file located at 'DPDN-Pytorch1.9.0-Cuda11.2/model/losses.py'.
In the source code:
def PoseDis(r1, t1, s1, r2, t2, s2):
'''
r1, r2: b33
t1, t2: b3
s1, s2: b3
'''
dis_r = torch.mean(torch.norm(r1 - r2, dim=1))
dis_t = torch.mean(torch.norm(t1 - t2, dim=1))
dis_s = torch.mean(torch.norm(s1 - s2, dim=1))

return dis_r + dis_t + dis_s

Both r1 and r2 are tensors with shape b33. Therefore, should dis_r = torch.mean(torch.norm(r1 - r2, dim=1)) be modified to dis_r = torch.mean(torch.norm(r1 - r2, dim=(1,2)))?