Stateful LSTM with Conv1D preprecessing in Tensorflow

Question

I currently have a stateful LSTM for sequence prediction in Tensorflow. My model is like this:

model = tf.keras.Sequential(