Does the current implementation allow for batch inference, is there an example / code how to do batch inference? Thx