I have a story with a form, where a user is asked to give more information (a course title) and then they story should continue (more actions). To keep it simpler, I shortend the story so that after the slot of the form has filled, there are two utters, one that prints out the slot.
However, the problem is, that after the form has ended (- active_loop: null) the story is not continued. Instead this is shown in the console Predicted next action 'action_listen' with confidence 0.93.
Can anybody help me out here? Is there something wrong with my config or the way I wrote the story with the form?
Thanks in advance!
I already deleted all the trained models and trained it again.
rasa --version:
Rasa Version : 2.8.1
Minimum Compatible Version: 2.8.0
Rasa SDK Version : 2.8.1
Rasa X Version : None
Python Version : 3.7.2
Operating System : Darwin-20.6.0-x86_64-i386-64bit
Python Path : /Users/theresa/.pyenv/versions/3.7.2/bin/python3.7
Thanks for the input. It’s true, it works with rules. However this story should end with a checkpoint which is not possible with rules, right? So I need to use stories
So I transform the story from above as a rule and just not use checkpoints, instead I just let the rule continue (with the actions from the checkpoints)? This is kind of an overhead but I think it would work. I could do that.
Nevertheless I am wondering why it’s not working with the story above? I think I used the form correctly (as stated in the documentation). I just like the usage of checkpoints as it makes (similar) stuff (with different a lot more clearer and better to maintain.
No, you keep all your current stories and checkpoints as they are. Just add the rule.
It’s just because stories don’t force the bot to act in some way, so they’re not always reliable.
I usually prefer to write short simple intent-to-action rules, then take stories from real conversations (of course they should be correct or manually corrected).
so I added the rule and updated the story to end with a checkpoint (to another story). However now the following is happening: Contradicting rules or stories found. Do I need to adapt the story further? Can I use the rule in the story?
As far as I understood the type “from_text”, it means any input given after the form has started, is the required_slot (in this case the current_course_title).
After the form was submitted, the checkpoint should be used (to trigger the next story) which needs this required slot to get some data from an API.