🧮 Fix max_steps calculation in RLOOTrainer

[qgallouedec]

What does this PR do?

Fixes #2401

I'm not entirely sure why it works but the problem is gone for varying batch size, rloo k, world size, grad accumulation step.

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? 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.
• 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.

[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.

[kashif]

is it because we have effectively 2 x bigger batches from the chosen and rejected pairs?

[qgallouedec]

Probably..., merging as it seems to resolve the issue.