Rasa bot for low resource languages

I am really new to Rasa and planning to use Rasa for a low resource/underrepresented language (Kinyarwanda). From my research, I think I need a Language Model for this language because neither Rasa nor Spacy has a pretrained Language model for this language.

My question is do I have to train a language model specifically and connect it somehow to Rasa?. Typically I don’t know where to start from.


Do you have the language in FastText - fastText/crawl-vectors.md at master · facebookresearch/fastText · GitHub

if so, you can take a look at vincent’s helper code for integrating word vectors from FastText https://rasahq.github.io/rasa-nlu-examples/docs/featurizer/fasttext/

No I don’t. Unfortunately.


the language kinarwanda exists as fastText vectors

Wow… Thanks a lot. I appreciate. Would look into this