Unet fix

[canberk17]

What does this PR do?

I was doing the CustomDiffusion tutorial, and I started receiving an UnboundLocalError at the end of the training.

File "/content/diffusers/examples/custom_diffusion/train_custom_diffusion.py", line 1330, in <module>
    main(args)
  File "/content/diffusers/examples/custom_diffusion/train_custom_diffusion.py", line 1312, in main
    images=images,
UnboundLocalError: local variable 'images' referenced before assignment

The line (1312) is referring inside the conditional block if args.push_to_hub:. This error error occurs because the images variable is defined inside two nested conditional blocks (lines 1232 and 1290), but is later referenced without being assigned if certain conditions are not met.

Here's how images is defined:

line 1232

if accelerator.is_main_process:
    if args.validation_prompt is not None and global_step % args.validation_steps == 0:
        # ...

line 1290

if args.validation_prompt and args.num_validation_images > 0:
    images = [
        # ...
    ]

This issue arises if you try to push your model to the hub without, setting the num_validation_images, and it will lead to a "UnboundLocalError" when the code tries to access images without having defined it first.

To ensure that the images variable is always initialized, we can declare an empty list at the beginning of the code block where the main process is checked. This will prevent the UnboundLocalError when you reference images when pushing your model to the hub, and don't define the num_validation_images

Referring to these: #4704 #4707

Before submitting

• This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
• Did you read the contributor guideline?
• Did you read our philosophy doc (important for complex PRs)?
• Was this discussed/approved via a Github issue or the forum? Please add a link to it if that's the case.
• Did you make sure to update the documentation with your changes? Here are the
  documentation guidelines, and
  here are tips on formatting docstrings.
• [] Did you write any new necessary tests?

Who can review?

@sayakpaul @patrickvonplaten

[sayakpaul]

Thanks!

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint.

[sayakpaul]

Will merge once the CI is green :-)

Thanks a lot for spotting out the bug and for your contribution.

[canberk17]

Would you like me to apply the black code formatter, and push it again?

[sayakpaul]

Once the CI is green, we can merge :-)