[tdoublep]

What does this PR do?

This fixes an issue for models like mistralai/Mamba-Codestral-7B-v0.1 that use mamba2 architecture with n_groups>1. It is equivalent to the fix for Zamba from #35943.

This issue prevents us comparing against transformers as baseline in vLLM CI for this model.

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?

@pglorio @vasqu @ArthurZucker @hmellor

[github-actions]

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

run-slow: bamba, falcon_h1, granitemoehybrid, mamba2

[vasqu]

Can we also let the zamba2 model inherit this from mamba2 then? Let's not introduce alternatives here

And could you add a fast test to mamba2 that might catch this?

[vasqu]

Suggested change

	def __init__(self, hidden_size, group_size, eps=1e-6):
	super().__init__(hidden_size, group_size, eps)
	pass

Weird that this even used the init here, we shouldn't need to anything on the modular side

[vasqu]

Same for the other modular codes

[ArthurZucker]

Thanks for the PR, and sorry it didn't make it in! This branch is based on v4, and v5 is very different, so rather than rebasing we'd love to invite you to open a new PR against main using our latest tools and good practices:

• Transformers tenets
• mlinter and the modeling rules it enforces (make typing)
• Modular transformers
• Contributing guide

Closing this one in the meantime, thanks again!