I want to know about test stories in RASA. I know that we can write test stories using RASA X using interactive learning. But if I generate a story like this, does it get saved in the bot memory? Because I want to know how my bot will perform end to end. So i have the real conversation from users. I am trying to convert that into the correct format for RASA test_stories. The data is huge so I want to know if I can automate the process of creating Rasa test_stories file other than using RASA X. and also if I use rasa x, and generate the story format from interactive learning, does it get saved in bot memory because i want the test results to be as they were on unseen data…
so i have two questions…
does the data i use to generate my test_stories.yml file using rasa x interactive learning get saved in the bot memory? because that will make the results of tests biased.
is there any way I can automate the process of creating stories for test_stories.yml file automatically?
Have a look at the Rasa server HTTP API. It has an endpoint to access a conversation’s end-to-end story that you can directly use as a test story. The workflow to build it is:
run Rasa shell in debug mode with API enabled
talk to the bot
copy the conversation id for the conversation above from the logs
use it in a request to the endpoint
It seems that the API only considers the internal tracker store to generate end-to-end stories (at least I could not get it working with an external tracker like PostgreSQL).