Rasa x fallback custom actions

Hello! So I have deployed Rasa X using the Docker Compose Installation. I have an action server running on a docker image. So I followed the documentation to override the docker-compose.yml to specify my docker image url.

I can see that my action server is running and connected to rasa x from the above screenshot attached.

My only problem here is that my action server overrides the 'action_default_fallback' (if you see the screenshot). I have written my own fallback that specifies to make an api call to get the required output. i.e dispatcher.utter_message(text=api_output)

The issue is when I type in the message using rasa x I keep getting the default fall back message and not the custom action message that I have written.

Can you please tell me what might I be missing here @stephens.

Thanks

Do you have the fallback classifier enabled in your pipeline and a fallback rule?

Yes @stephens this is how my config file looks like, language: en pipeline:

  • name: WhitespaceTokenizer
  • name: RegexFeaturizer
  • name: LexicalSyntacticFeaturizer
  • name: CountVectorsFeaturizer
  • name: CountVectorsFeaturizer analyzer: char_wb min_ngram: 1 max_ngram: 4
  • name: DIETClassifier epochs: 100
  • name: EntitySynonymMapper
  • name: ResponseSelector epochs: 100
  • name: FallbackClassifier threshold: 0.4 policies:
  • name: MemoizationPolicy
  • name: TEDPolicy max_history: 5 epochs: 100
  • name: RulePolicy

This is the response i’m getting. I know its going to action_default_fallback but the response is always empty.

Fixed the problem. I was missing to include the actions_default_feedback in the domain. Assumed it to auto include for just like any other default domain.yml response. @stephens

Hello @stephens… What seemed to work locally doesn’t seems to be working on Rasa X. I am doing the docker compose installation. It just doesn’t seem to work… Can you help me a bit here?

Thanks

Can you try the rest endpoint for the server? Are you sure it’s the same model? Do you have a test story that is working?

Hey @stephens ,

Thanks for the response… It was an environment and the way I had set up my Rasa X environment that was causing the issue I guess. There are a lot of changes with Rasa 2.0. While I was reading the documentation I also referred to the masterclass video. So there was a lot of conflict in the configuration of Rasa X. The issue is fixed. Thanks for your time.