Add eurobert (#39455)
* feat: init EuroBert model integration
* fix: auto import
* Refactor: Update EuroBertMaskedLM & SentenceClassif classes to inherit from PreTrainModel.
* feat: Introduce EuroBertPreTrainedModel class and update model inheritance
* fix: Ruff format
* fix: Ruff format for modeling_auto.py
* fix: docstrings of public objects.
* Fix: EuroBERT config doc
* fix: Update classifier_pooling documentation and ensure consistent default value
* fix: ruff format
* feat: EuroBERT Doc file creat
* fix: use_cache and pretraining_tp params in config, uneeded docstring information and typo in docs
* test: Ported Llama unit tests to EuroBert
* fix: fixed test file name, removed unnecessary attention mask creation and remove unneeded comment
* fix: run ruff formatting
* docs: initial documentation for eurobert model.
* fix: update documentation to reflect EuroBERT model and adjust attention mask condition
* fix: remove unnecessary else clause for attention mask in EuroBert model
* fix: correct attention mask handling in EuroBert model
* fix: Same but on right file this time :p
* feat: add EuroBertForTokenClassification model for NER tasks and update related documentation
* remove
* revert wrong merge
* oups
* modular fix
* simplify and refactor the modular
* more fixes
* Update and modernize EuroBERT architecture; patch & add tests; docs
* Remove use_cache from config, anything other than False isn't allowed
* Use ModernBERT-style docs float
* Rerun add_dates
* Rerun add_dates.py again
---------
Co-authored-by: NICOLAS BZRD <79255399+Nicolas-BZRD@users.noreply.github.com>
Co-authored-by: DuarteMRAlves <duartemalves@tecnico.ulisboa.pt>
Co-authored-by: Tom Aarsen <Cubiegamedev@gmail.com>
Co-authored-by: Tom Aarsen <37621491+tomaarsen@users.noreply.github.com>
