Fix embeddings resizing in TF models (#8657)

* Resize the biases in same time than the embeddings

* Trigger CI

* Biases are not reset anymore

* Remove get_output_embeddings + better LM model detection in generation utils

* Apply style

* First test on BERT

* Update docstring + new name

* Apply the new resizing logic to all the models

* fix tests

* Apply style

* Update the template

* Fix naming

* Fix naming

* Apply style

* Apply style

* Remove unused import

* Revert get_output_embeddings

* Trigger CI

* Update num parameters

* Restore get_output_embeddings in TFPretrainedModel and add comments

* Style

* Add decoder resizing

* Style

* Fix tests

* Separate bias and decoder resize

* Fix tests

* Fix tests

* Apply style

* Add bias resizing in MPNet

* Trigger CI

* Apply style

templates/adding_a_new_model/cookiecutter-template-{{cookiecutter.modelname}}/modeling_tf_{{cookiecutter.lowercase_modelname}}.py

@@ -772,10 +772,16 @@ class TF{{cookiecutter.camelcase_modelname}}ForMaskedLM(TF{{cookiecutter.camelca
 
         self.{{cookiecutter.lowercase_modelname}} = TF{{cookiecutter.camelcase_modelname}}MainLayer(config, name="{{cookiecutter.lowercase_modelname}}")
         self.mlm = TF{{cookiecutter.camelcase_modelname}}MLMHead(config, self.{{cookiecutter.lowercase_modelname}}.embeddings, name="mlm___cls")
-
+    
     def get_output_embeddings(self):
         return self.{{cookiecutter.lowercase_modelname}}.embeddings
 
+    def get_output_layer_with_bias(self):
+        return self.mlm.predictions
+
+    def get_prefix_bias_name(self):
+        return self.name + "/" + self.mlm.name + "/" + self.mlm.predictions.name
+
     @add_start_docstrings_to_model_forward({{cookiecutter.uppercase_modelname}}_INPUTS_DOCSTRING.format("batch_size, sequence_length"))
     @add_code_sample_docstrings(
         tokenizer_class=_TOKENIZER_FOR_DOC,