What's next after starter-pack

Hi there. I am still a newbie of RASA chatbot. I have done all things here https://github.com/RasaHQ/starter-pack-rasa-stack with my chatbot telling a Chuck Norris jokes now. But I don’t know what’s next.

Clearly this should be the answer. I have appended the smalltalk.md to data/nlu_data.md and retrained RASA NLU with it. But it turned out my chatbot is still a stupid bot telling jokes all the time, without new skills learned. What’s the problem? and what other things could be done?

Now I am trying to assess the Messaging App.

Hwy @Neptune. Great to have you here and great to see you started with the starter-pack! After you append smalltalk.md to the original training examples, you teach your assistant to understand more possible user inputs. The task for you is to enrich the stories.md data with stories which would include these new intents, and add them to domain.yml alongside new actions and templates. Once you do that, your assistant should start picking up new skils you define.

Thanks, Juste! You are saying I need to write down customized stories myself based on newly appended intents? Am I getting you the right way? Also, after I enrich the stroies.md, I need to write downs actions based on newly appended intents, right?:sunglasses:

Yes, exactly. You can use the newly appended intents to think of what new skills could invlolve them,write the stories in stories.md accordingly; then implement those skills as templates inside domain.yml or as cutom actions inside the actions.py. :slightly_smiling_face:

Cool! Thank you Juste!:sunglasses:

Hi, i am trying to amend the stories.md file to define a story with a default welcome text to begin the chat

this is one story i am writing

story_opening

  • greet
  • utter_salute
  • action_joke

i think without any (*greet) message from user can i directly start the chat session with a welcome message like

story_opening

  • utter_salute
  • utter_abilities

Thanks

Hi,

I stumbled upon this thread later

and the response

will look into the workaround Thanks :slight_smile: