Keras clarification of using preprocessing with tf.data with multiple backend

Say, I have a preprocessing layer, i.e. image resize or any augmentation methods, build with keras.ops. And I use it with tf.data API. Now, with another backend, i.e. torch/jax, I can train the model with tf.data API. Now, having say jax/torch as backend and keras.ops with tf.data api - will it work with model.fit? As using keras.ops with jax/torch backend will call respected operations, will tf.data with those preprocessing layer work as expected?