Fix fast tokenization problems (#13930)

* Fix albert mask token tokenization.

* Ensure special tokans sanitized.

* Style

* Fix

* Apply suggestions from code review

src/transformers/models/albert/tokenization_albert.py

@@ -142,8 +142,13 @@ class AlbertTokenizer(PreTrainedTokenizer):
         sp_model_kwargs: Optional[Dict[str, Any]] = None,
         **kwargs
     ) -> None:
-        # Mask token behave like a normal word, i.e. include the space before it
-        mask_token = AddedToken(mask_token, lstrip=True, rstrip=False) if isinstance(mask_token, str) else mask_token
+        # Mask token behave like a normal word, i.e. include the space before it and
+        # is included in the raw text, there should be a match in a non-normalized sentence.
+        mask_token = (
+            AddedToken(mask_token, lstrip=True, rstrip=False, normalized=False)
+            if isinstance(mask_token, str)
+            else mask_token
+        )
 
         self.sp_model_kwargs = {} if sp_model_kwargs is None else sp_model_kwargs
 

src/transformers/models/albert/tokenization_albert_fast.py

@@ -135,8 +135,13 @@ class AlbertTokenizerFast(PreTrainedTokenizerFast):
         mask_token="[MASK]",
         **kwargs
     ):
-        # Mask token behave like a normal word, i.e. include the space before it
-        mask_token = AddedToken(mask_token, lstrip=True, rstrip=False) if isinstance(mask_token, str) else mask_token
+        # Mask token behave like a normal word, i.e. include the space before it and
+        # is included in the raw text, there should be a match in a non-normalized sentence.
+        mask_token = (
+            AddedToken(mask_token, lstrip=True, rstrip=False, normalized=False)
+            if isinstance(mask_token, str)
+            else mask_token
+        )
 
         super().__init__(
             vocab_file,