[fix] Use the correct _tied_weights_keys for CamembertForCausalLM

[tomaarsen]

What does this PR do?

Follow-up for #44931, which added weight tying for Camembert.
Only the CamembertForMaskedLM class had the right _tied_weights_keys, the CamembertForCausalLM had the incorrect weight key.

Details

from transformers import AutoModelForCausalLM

model = AutoModelForCausalLM.from_pretrained("hf-internal-testing/tiny-random-camembert")
print(type(model))

On main:

Traceback (most recent call last):
  File "[sic]\demo_camembert.py", line 4, in <module>
    model = AutoModelForCausalLM.from_pretrained("hf-internal-testing/tiny-random-camembert")
            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "[sic]\src\transformers\models\auto\auto_factory.py", line 381, in from_pretrained
    return model_class.from_pretrained(
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "[sic]\src\transformers\modeling_utils.py", line 4087, in from_pretrained
    model = cls(config, *model_args, **model_kwargs)
            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "[sic]\src\transformers\models\camembert\modeling_camembert.py", line 1171, in __init__
    self.post_init()
  File "[sic]\src\transformers\modeling_utils.py", line 1298, in post_init   
    self.all_tied_weights_keys = self.get_expanded_tied_weights_keys(all_submodels=False)   
                                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^   
  File "[sic]\src\transformers\modeling_utils.py", line 2490, in get_expanded_tied_weights_keys
    raise ValueError(
ValueError: There is an issue with your definition of `tie_weights_keys` for ^camembert.embeddings.word_embeddings.weight:^lm_head.decoder.weight. We found [] to tie into ['lm_head.decoder.weight']

On this PR:

<class 'transformers.models.camembert.modeling_camembert.CamembertForCausalLM'>

Code Agent Policy

• I confirm that this is not a pure code agent PR.

I wrote the code and PR description myself

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?

@ArthurZucker @Cyrilvallez

• Tom Aarsen

[github-actions]

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

run-slow: camembert

[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.

[Cyrilvallez]

Oh damn indeed! Thanks!