Update logits_process.py docstrings to clarify penalty and reward cases (attempt #2)

[larekrow]

This PR fixes point 3 of #25970 by clarifying the penalty and reward cases for RepetitionPenaltyLogitsProcessor and EncoderRepetitionPenaltyLogitsProcessor within the docstrings.

PR #26129 was the original copy, but I have accidentally deleted my repo that submitted the PR, so I cannot reopen that PR 🙇

@gante, for your review please.

[ArthurZucker]

LGTM thanks 😉

[ArthurZucker]

Could you run make fixup to pass the failing test

[larekrow]

Hey @ArthurZucker, I tried to run make fixup but encountered an error in repo-consistency.

Traceback (most recent call last):
  File "/home/users/user/temp/transformers/utils/update_metadata.py", line 337, in <module>
    check_pipeline_tags()
  File "/home/users/user/temp/transformers/utils/update_metadata.py", line 316, in check_pipeline_tags
    model = model[0]
IndexError: tuple index out of range
make: *** [Makefile:44: repo-consistency] Error 1

Anyway I thought I should not be encountering this in the first place since I only modified docstrings in a single file. So, I ran black src/transformers/generation/logits_process.py which gave:

All done! ✨ 🍰 ✨
1 file left unchanged.

Even ruff src/transformers/generation/logits_process.py --fix does nothing. Not sure what I should do to pass the failing test? I'm using black==23.9.1 and ruff==0.0.292 if that helps.

[ArthurZucker]

Hey! we have "ruff>=0.0.241,<=0.0.259" so the version you are using is probably too far! Try downgrading or doing something likepip install -e ".[quality]"and then runmake style`

[larekrow]

Thanks for the guidance @ArthurZucker! The test has passed 😄

[ArthurZucker]

Cool thanks for the contribution

[HuggingFaceDocBuilderDev]

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