I had build the hybrid CNN and LSTM model using Tensorflow 2.17 version. I want to add the backpropagating tuning to the LSTM to modify the weight of the LSTM layers. I try to find a way to access the weight to do calculation and update the weight during training for the LSTM . I had do some research where they build the LSTM manually instead apply tensorflow API. I just curious if there is a way that I could access the weight , then apply derivative and finally update while training.

Here is my model :

    model_cnn = tf.keras.models.Sequential([
        tf.keras.layers.Conv1D(input_shape = input_data_shape,
                               filters=256,
                               kernel_size=3,
                               activation='linear',
                               strides=1,
                               padding="causal"),
        tf.keras.layers.MaxPooling1D(pool_size=2),
        tf.keras.layers.Conv1D(filters=128,
                               kernel_size=3,
                               activation='linear',
                               strides=1,
                               padding="causal"),
        tf.keras.layers.MaxPooling1D(pool_size=2),    
        tf.keras.layers.Dropout(0.5),
        tf.keras.layers.LSTM(16 ,return_sequences=True),
        tf.keras.layers.LSTM(8,return_sequences=True),
        tf.keras.layers.Flatten(),
        tf.keras.layers.Dense(1)
    ])
    return model_cnn

My expectation is to access the weight and update the weight during training of the hybrid CNN and LSTM model.