Hi,
I’m using the ResponseSelector component in my NLU pipeline for handling faqs/chichat. Model training works fine locally, however, seems that RasaX doesn’t recognise/use the retrieval intent responses you define withing the NLU files if you train a model through its UI. For example, I’m getting the following warnings in the logs when using my bot’s model trained through RasaX:
rasa-production_1 | /opt/venv/lib/python3.7/site-packages/rasa/shared/utils/io.py:93: UserWarning: Unable to fetch response templates for faq/helpdesk_availability This means that there is likely an issue with the training data. Please make sure you have added response templates for this intent.
rasa-production_1 | /opt/venv/lib/python3.7/site-packages/rasa/shared/utils/io.py:93: UserWarning: Unable to fetch response templates for chitchat/bot_challenge This means that there is likely an issue with the training data. Please make sure you have added response templates for this intent.
rasa-production_1 | /opt/venv/lib/python3.7/site-packages/rasa/shared/utils/io.py:93: UserWarning: Unable to fetch response templates for chitchat/how_are_you This means that there is likely an issue with the training data. Please make sure you have added response templates for this intent.
rasa-production_1 | 2020-11-18 23:13:47 ERROR rasa.core.actions.action - Couldn't create message for response 'utter_faq/helpdesk_availability'.
This issue no longer appears if I move these responses to the domain.yml file, however, I’m getting a Project validation completed with errors message when running the rasa data validate command, as I described here