revert behavior of _prepare_from_posids

[winglian]

What does this PR do?

#39474 causes the grad norm to become NaN during training. We don't need to revert the whole thing, but reverting just this function seems to fix the regression.

@ArthurZucker @SunMarc

pre-#39474 loss

post-#39474 loss

with this fix

Fixes # (issue)

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.

[ArthurZucker]

Thanks, this won't work for inference! can we just have a if else!

[ArthurZucker]

(Thanks a lot for checking sorry for breaking)

[ArthurZucker]

Thanks makes sense!