for this update you can use python parser1.py --len=? to determine how long each model contain if use 3, it will be tourist tourist tourist guide guide guide (current speaker) next, you can use python intonly.py to run intent now it intonly.py only train using last sentence