Pause story and continue where you left off

Great! Being able to parse all the data --debug sends at you is an important skill in using Rasa,

Too bad it doesn’t help with your original issue. This just tells me you don’t have a story that goes

  • faq_sports
    • utter_sports (and utter_more_questions or sum’n)
  • faq_dreamteam
    • utter_dreamteam

the NLU engine does pick up the differences though. That’s good! Are you using SpaCY-nl or supervised_embeddings?

Now to help you further, what version of Rasa_core are you running?

I don’t think we are using SpaCY-nl, but we do use supervise_embeddings.

If I’m correct, I am running Rasa Core 0.14.0. Rasa just launched a new version right? Should I update Rasa? And will this keep my current progress intact?

Ah okay, rasa-core 0.14.1 fixed a bug with MappingPolicy:

[0.14.1] - 2019-05-02

Fixed

  • fixed MappingPolicy bug upon prediction of ACTION_LISTEN after mapped action

You can update to 0.14.3 with no problems. I don’t know what happens if you update to 1.0.1, because that changes a LOT of stuff.

Thanks! I hoped updating would make it all work… I updated Rasa to 0.14.3, but it still does something weird…

output5.txt (12.8 KB)

So it actually answers the FAQ, then I answer the bot’s question, but then it gives me two utters on which I should answer one too (what year I’m starting).

Edit

Updated to 0.14.4, other behavior.

output6.txt (12.8 KB)

It now assumes the question is also an answer to a question that wasn’t even asked yet… :disappointed: Telling when the deadline is should be after giving an answer to “When are you starting?” “this year”, but it seems to just assume the answer is ‘this year’

Anybody who can still help me?

Hi Nikki,

I see my last reply didn’t come through. My apologies! I’ve asked around in other threads. They’re working on a way to make MappingPolicy more easy to use.

For now, their solution is to make the utters you map to the intents into custom actions. You acn also just wait for their new solution.

I’ll link the thread in my next comment.

Okay thanks! I’ll try to make that work. Too bad it’s not easy to do right now. My project will end in a few weeks and will then be used the upcoming years on a public school website, so the chatbot will only have the functionality that is available at this very moment.

But still, thank you for your time!