Rules for training via HTTP server

Hello everyone! Does anybody knows if the HTTP API endpoint /model/train can receive an “rules” attribute to pass information for the training? Or if it accepts this information inside other attribute, like “nlu”?

I couldn’t find this information in the rasa api documentation… Rasa Open Source Documentation

I use Rasa http server to do all my training and I’m wondering how to pass the ‘rules.yml’ information to the HTTP server. Thx!!

i also wondering why this was completely forgotten in migration. As i can see there is no consideration in JSON training data for rules. JSON will be handled as markdown as before. maybe there is a possibility to provide rules data in story key. This will be handled as training files. The docs are not very helpful as there is missing yaml example. Maybe it is possible to provide rules in the yaml string and set application/x-yaml header. You can also write your stories as before:

An example of a rule where the bot returns a fixed response “utter_greet” to a user message with intent “greet” would be: rules:

  • rule: Greeting Rule steps:
    • intent: greet
    • action: utter_greet

For multiple-turn interactions, you should define a story, for example: stories:

  • story: Greeting and ask user how they’re doing steps:
    • intent: greet
    • action: utter_greet
    • action: utter_ask_how_doing
    • intent: doing_great
    • action: utter_happy
2 Likes

Hi @Tobias_Frank,

I came across this issue aswell and decided to open a Github issue for it.

I would be willing to fix this / contribute the missing functions.

Kind regards
Julian