Running diffusers with torch-DDP leads to warning on stride mismatch

[jinc7461]

Describe the bug

This issue has already been reported at #3809 Through an analysis of the mismatched network parameters, it was found that modifying line 369 in diffusers/models/resnet.py from:

input_tensor = self.conv_shortcut(input_tensor)

to

input_tensor = self.conv_shortcut(input_tensor.contiguous())

resolves the problem.

Reproduction

This issue has already been reported at #3809.

Logs

System Info

Diffusers: diffusers==0.31.0
Torch: torch==2.6.0
Python: Python 3.10.15

Who can help?

@DN6 @yiyixuxu @sayakpaul

[hlky]

Hi @jinc7461. Thanks for raising this and providing the fix, would you like to open a PR?