Model does not return new intents

[agerten (Anne)]

Hi, I am using Python 3.8.6 and installed rasa-2.1.3.

I didn’t change anything on the default configuration, I just trained the model with the bootstrap data and ran the Rasa shell command to test the chatbot. The chatbot returns results fine.

Now I added one more intent “weather” to the nlu.yml file and modified the domain.yml as well:

The model was retrained successfully, but it does not seem to pick up an answer for the new intent “weather”:

The answer is either empty or falls back to another intent, if I ask the exact question from the intent defined questions.

More information running it in debug mode:

Can anybody advise?

Thank you!

[marcos.allysson]

@agerten, you should add a story on your data/stories.yml file including this new path of conversation to your assistant. Retrain your model and try again. Hope it helps . You can even run rasa interactive to teach your bot handle this new path.

[agerten (Anne)]

Thank you! Just to confirm, I assumed just matching in the response name utter_weather with the intent name weather would indicate that this intent should be used. But if I understand right, it is required to map each intent to its response via a story?

Thanks again!

[marcos.allysson]

you don’t necessarily need to match your intent name with response. You can clearly create an intent “greet”, with a utter_intro response name… You just have to create a storie on stories.yml to teach your assistant to answer a certain intent with a response. Similar to a real conversation, if someone says “hi” to you, you’d respond back with a greeting. Someone asks your name, you’d do the same. Just like stories. Got it @agerten ?