Add Fast Image Processor for ImageGPT

[agamjots05]

What does this PR do?

Related #36978

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,
  Pull Request section?
• 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?

Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.

[Rocketknight1]

cc @yonigozlan

[yonigozlan]

Hey @agamjots05 and @ayaayethan ! Thanks for opening this PR, don't hesitate to ping me when it's ready to review or if you need help!

[agamjots05]

Hey @agamjots05 and @ayaayethan ! Thanks for opening this PR, don't hesitate to ping me when it's ready to review or if you need help!

Thank you for letting us know! We'll let you know if we get stuck on anything 👍

[agamjots05]

@yonigozlan Quick question
My team and I have made the fast processor for ImageGPT and have come across an interesting issue in regards to the testing file for ImageGPT.

To get the final two slow/fast parity tests to pass, we needed to make two very small changes in the slow ImageGPTImageProcessor:

1. Always include pixel_values in the returned BatchFeature even when do_color_quantize=True (alongside input_ids).
   This avoids KeyError/AttributeError in the shared test mixins that access encoding.pixel_values. It’s additive-only.
2. Override to_dict() to convert clusters from np.ndarray to a plain Python list.
   This makes the config JSON-serializable, which allows us to pass a test we were failing

Would these two minimal adjustments to the slow processor be acceptable?
If not should we, for the time being, ignore these failing test cases occurring due to the base processor?

[yonigozlan]

Hey @agamjots05 !
These seems like reasonable changes, strange that it didn't cause any issues before, thanks for fixing!

[agamjots05]

Hey @agamjots05 ! These seems like reasonable changes, strange that it didn't cause any issues before, thanks for fixing!

Sounds good. If that's the case, then this PR should be ready for review 👍

[yonigozlan]

Hey @agamjots05 and @ayaayethan , thanks for contributing! There's a few changes to do, and I'll be off for two weeks, but happy to do another review when I'm back!

[agamjots05]

@yonigozlan I believe I have implemented all the changes you have requested. Please take a look when you can 👍

[github-actions]

[For maintainers] Suggested jobs to run (before merge)

run-slow: auto, imagegpt

[yonigozlan]

Thanks for working on this @agamjots05 and @ayaayethan !
I just pushed some small simplifications to the color quantize logic, and I'll merge once the CI passes!

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.