Use pretrained spacy Embeddings from project directory

Rasa Open Source
1

Hi, I want to prototype a rasa NLU model. To set it up I use pre-trained spacy embeddings. I had to initially do:

python -m spacy download de_core_news_md
python -m spacy link de_core_news_md de

To set it up. Now I start the NLU server with:

rasa run --enable-api -m models/nlu-model.tar.gz

I observed that my model is loading the spacy embeddings from wherever spacy has put the embeddings in the first step. Is it possible to set up my project directory in a way that everything the model needs to run is inside the project directory? How can I do it?

Thank you very much

2

Unfortunately the spacy CLI doesn’t seem to provide a way to give a path to where you want your file downloaded or the symlink created.