ValuError: Exception encountered when calling tf-distil_bert_model

Question

When i load my jupyter notebook to Google colab i get an error, but on my laptop I do not get this error. the error is :

ValueError: layer Exception encountered when calling 'tf_distil_bert_model' (type TFDistilBertModel).

Data of type <class 'keras.src.engine.keras_tensor.KerasTensor'> is not allowed only (<class 'tensorflow.python.framework.tensor.Tensor'>, <class 'bool'>, <class 'int'>, <class 'transformers.utils.generic.ModelOutput'>, <class 'tuple'>, <class 'list'>, <class 'dict'>, <class 'numpy.ndarray'>) is accepted for input_ids.

Call arguments received by layer 'tf_distil_bert_model' (type TFDistilBertModel): input_ids=["<KerasTensor: shape=(None, 256) dtype=int32 (created by layer 'input_ids')>", "<KerasTensor: shape=(None, 256) dtype=int32 (created by layer 'input_attention')>"] attention_mask=None head_mask=None inputs_embeds=None output_attentions=None output_hidden_states=None return_dict=None training=False

Anyone knows why i get this error in colab? I was building a pre-trained DISTILBERT model with some additional tensorflow layers as shown below:

MAX_LENGTH = 256
LAYER_DROPOUT = 0.5
LEARNING_RATE = 1e-4
RANDOM_STATE = 42

def build_model(transformer, max_length=MAX_LENGTH):
    # Define weight initializer with a random seed to ensure reproducibility
    weight_initializer = tf.keras.initializers.GlorotNormal(seed=RANDOM_STATE) 

    # Define input layers
    input_ids_layer = tf.keras.layers.Input(shape=(max_length,), 
                                        name='input_ids', 
                                        dtype='int32')
    input_attention_layer = tf.keras.layers.Input(shape=(max_length,), `your text`
                                              name='input_attention', 
                                              dtype='int32')

    # DistilBERT outputs a tuple where the first element at index 0
    # represents the hidden-state at the output of the model's last layer.
    # It is a tf.Tensor of shape (batch_size, sequence_length, hidden_size=768).
    last_hidden_state = transformer([input_ids_layer, input_attention_layer])[0]

    # We only care about DistilBERT's output for the [CLS] token, 
    # which is located at index 0 of every encoded sequence.  
    # Splicing out the [CLS] tokens gives us 2D data.
    cls_token = last_hidden_state[:, 0, :]

    dense_layer1 = tf.keras.layers.Dense(256, 
                               activation='relu',
                               kernel_initializer=weight_initializer,  
                               kernel_constraint=None,
                               bias_initializer='zeros'
                               )(cls_token)

    dropout_layer1=tf.keras.layers.Dropout(LAYER_DROPOUT)(dense_layer1)
    dense_layer2 = tf.keras.layers.Dense(256, 
                               activation='relu',
                               kernel_initializer=weight_initializer,  
                               kernel_constraint=None,
                               bias_initializer='zeros'
                               )(dropout_layer1)

    dropout_layer2=tf.keras.layers.Dropout(LAYER_DROPOUT)(dense_layer2)
    dense_layer3 = tf.keras.layers.Dense(32, 
                               activation='relu',
                               kernel_initializer=weight_initializer,  
                               kernel_constraint=None,
                               bias_initializer='zeros'
                               )(dropout_layer2)

    dropout_layer3=tf.keras.layers.Dropout(LAYER_DROPOUT)(dense_layer3)

    # Define a single node that makes up the output layer (for binary classification)
    output = tf.keras.layers.Dense(1, 
                               activation='sigmoid',
                               kernel_initializer=weight_initializer,  
                               kernel_constraint=None,
                               bias_initializer='zeros'
                               )(dropout_layer3)

    # Define the model   
    model = tf.keras.Model([input_ids_layer, input_attention_layer], output)

    # Compile the model
    model.compile(tf.keras.optimizers.Adam(learning_rate=LEARNING_RATE), 
              #loss='binary_crossentropy',
              loss=tf.keras.losses.BinaryFocalCrossentropy(),
              metrics=['accuracy'])
    return model 

model=build_model(distilBERT)

Answer 1

I turns out that the reason for the error is because the version of transformers module on my laptop and that on colab were different. So now i install the same version of transformers as I have on my laptop and that fixed the problem.