Dear authors,
Your paper "Faster Attend-Infer-Repeat with Tractable Probabilistic Models" is really interesting and i like that so much. However, there is still a question in my mind, that is, how to use the proposed method to generate new scenes? Since there isn't any decoder like stuff to assist the model to do that as far as i understand.
So, sincerely to looking for your replies!
Thanks for your work again!