Bert Istantiation TypeError: 'NoneType' object is not callable Tensorflow

Question

I'm trying to instantiate a BERT model using TensorFlow. This code was running correctly until a few days ago, but now I'm getting this error. I've used this same code to instantiate BERT models on Kaggle in other notebooks and it worked fine. Can anyone give me a hint? I should note that I'm using Transformers version 4.31.0 to fix another error. The code I use is:

# Creazione del modello BERT per BYTECODE
BC_input_layer = Input(shape=(max_seq_length,), dtype=tf.int32, name="input_ids")
BC_bert_model = TFBertModel.from_pretrained("bert-base-uncased")
BC_pooler_output = BC_bert_model(BC_input_layer)[1]

BC_dropout_layer = Dropout(dropout_rate)(BC_pooler_output)
BC_output_layer = Dense(6, activation='sigmoid')(BC_dropout_layer)

BC_model = Model(inputs=BC_input_layer, outputs=BC_output_layer)

The error is

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
Cell In[10], line 19
     17 SC_input_layer = Input(shape=(max_seq_length,), dtype=tf.int32, name="input_ids")
     18 SC_mask_layer = Input(shape=(max_seq_length,), dtype=tf.int32, name="attention_mask")
---> 19 SC_bert_model = TFBertModel.from_pretrained("bert-base-uncased")
     20 SC_pooler_output = SC_bert_model(SC_input_layer, attention_mask=SC_mask_layer)[1]  # Estrai il secondo output, che è il pooler_output
     22 # Aggiungi un layer di Dropout

File /opt/conda/lib/python3.10/site-packages/transformers/modeling_tf_utils.py:2894, in TFPreTrainedModel.from_pretrained(cls, pretrained_model_name_or_path, config, cache_dir, ignore_mismatched_sizes, force_download, local_files_only, token, revision, *model_args, **kwargs)
   2892         model.build()  # build the network with dummy inputs
   2893 else:
-> 2894     model.build()  # build the network with dummy inputs
   2896 if safetensors_from_pt:
   2897     from .modeling_tf_pytorch_utils import load_pytorch_state_dict_in_tf2_model

File /opt/conda/lib/python3.10/site-packages/keras/src/layers/layer.py:224, in Layer.__new__.<locals>.build_wrapper(*args, **kwargs)
    221 @wraps(original_build_method)
    222 def build_wrapper(*args, **kwargs):
    223     with backend.name_scope(obj.name, caller=obj):
--> 224         original_build_method(*args, **kwargs)
    225     # Record build config.
    226     signature = inspect.signature(original_build_method)

File /opt/conda/lib/python3.10/site-packages/transformers/modeling_tf_utils.py:1131, in TFPreTrainedModel.build(self, input_shape)
   1129 def build(self, input_shape=None):
   1130     call_context = get_call_context_function()
-> 1131     if self.built or call_context().in_call:
   1132         self.built = True
   1133     else:

TypeError: 'NoneType' object is not callable