Warn the user about max_len being on the path to be deprecated. (#4528)

* Warn the user about max_len being on the path to be deprecated.

* Ensure better backward compatibility when max_len is provided to a tokenizer.

* Make sure to override the parameter and not the actual instance value.

* Format & quality

src/transformers/tokenization_utils.py

@@ -22,6 +22,7 @@ import logging
 import operator
 import os
 import re
+import warnings
 from collections import UserDict, defaultdict
 from contextlib import contextmanager
 from typing import Any, Dict, List, NamedTuple, Optional, Sequence, Tuple, Union
@@ -822,7 +823,14 @@ class PreTrainedTokenizer(SpecialTokensMixin):
         super().__init__(**kwargs)
 
         # For backward compatibility we fallback to set model_max_length from max_len if provided
-        model_max_length = model_max_length if model_max_length is not None else kwargs.pop("max_len", None)
+        if "max_len" in kwargs:
+            warnings.warn(
+                "Parameter max_len is deprecated and will be removed in a future release. "
+                "Use model_max_length instead.",
+                category=FutureWarning,
+            )
+
+            model_max_length = kwargs.pop("max_len")
         self.model_max_length = model_max_length if model_max_length is not None else VERY_LARGE_INTEGER
 
         # Padding side is right by default and overridden in subclasses. If specified in the kwargs, it is changed.