Word embeddings and RASA NLU

How to add fastText pre-trained/customized word embedding in RASA NLU? I’m trying to make a chatbot for ‘Bangla’ Language.

1 Like

Hey @nurakib. I think this thread will be helpful for you: Feeding Custom/Pretrained embeddings for ner_crf. Our contributor Souvik wrote a great tutorial on how to use fastText with Rasa, have a look and let us know if you need more help.

This works slightly more simply than that post now.

Once you convert the fastText model to spacy vectors, you can just add text_dense_features under CRFEntityExtractor's features, and your SpacyFeaturizer will add a dense vector per token.

The article in the medium about using fasttext is quite confusing, can I get some steps to do the same for other languages than English?.

Hi @Shrivatsa2904. I would suggest checking out one of our latest projects - Rasa NLU examples where you can also find components that support other languages as well.

Thank you, I’ll check them out.