Model does not know when to stop generating text?

#93
by jashsayani - opened

I am trying out this Falcon-7b model on my M1 Pro mac and noticed that it does not stop generating text. It keeps going till it reaches max_length. Is this an issue with the model? or some config that I am not setting correctly?

My pipeline (shows it generated the response twice; also includes prompt question in response but I think thats just how its going to generate response):
Screenshot 2023-11-05 at 8.34.05 PM.png

Sign up or log in to comment