Long story and short story execution issue

Hi,

I am a having a problem with multiple story execution within the same session. Once I execute a long story followed by a short one, the TED policy confuses with the responses.

Also even if I define the intent and following action as rules, the rule policy never predicts anything.

What parameters I need to choose at policy definitions to work with long stories? What should be the value of max_memory and epochs?

Have you run rasa data validate against your training data to see if it identifies story/rule conflicts?

Do you also have test stories that duplicate this issue?

Also, run rasa shell --debug to see details on the prediction issue and share it here.