Has this happened to others before? I have one entity that I need to fill from a lookup table since it’s categorical. First I thought I’d forgotten to add the lookup table or accidentally deleted it. Now I see that after using and training with Rasa-X, which has access to my YAML files, the lookup table keeps getting deleted even though I am using the Regex Entity Extractor and Regex Featurizer in my pipeline.
- name: RegexEntityExtractor
# use lookup tables to extract entities
use_lookup_tables: True
# use match word boundaries for lookup table
use_word_boundaries: False
I had been designing my conversations using Rasa and just recently started using Rasa-X. I never added lookup tables using the Rasa X UI. They had been added on my code editor instead, even the second time when they were gone and I realised there were missing, after which I restarted Rasa-X.
Update, I kind of realized that I had to manually go into Training > NLU data > Regex & Lookup and then choose to add the lookup table. Otherwise, it seems to get deleted from my NLU YAML file by Rasa-X. I think it’s a bit counter-intuitive.