Dialogue policy learning through RL

Hello, I hope you all are doing well. Thank you for organizing this forum for us to figure out the problems we are facing collectively. I plan to use RASA for dialogue modeling. However, I was the action policy to learned by an RL agent. I was hoping if you can guide me if RASA provides such support and if yes, then how can these customizations be done.

Thanks in advance. Looking forward to a response.

Hi ifrahidrees. Rasa currently uses supervised learning for the action policy, and does not yet supply a reinforcement learning framework.

1 Like