Hello, i am using Rasa 1.4.3 .
I have used fastText pretrained word embeddings to feed crf_entity_extractor with ner_features.
I have 200 dimensional vectors for word. Is it good idea to use ner_features this way?
The current version (which will be old in a week) of rasa is 1.7.2. My first advice is to upgrade.
Second … if the setup you have works well on your data. Then I can’t argue against it. Without knowing a little bit more about your bot/problem though, it is hard for me to give meaningful advice on what might possibly work better.