This project is intended for training Danish, Swedish and Norwegian sentence transformers. The project is an extension of the Danish Foundation models project.
You can install scandinavian-sentence-transformers
via pip:
git clone {repo url}
cd scandinavian-sentence-transformers
pip install -e .
but we recommend using invoke for the setup:
git clone {repo url}
cd scandinavian-sentence-transformers
# install invoke
pip install invoke
# setup up virtual environment and install dependencies
inv setup
To train the models you wi
inv prepare_dataset --lang da
inv train --model_name vesteinn/DanskBERT
pip install scandeval
scandeval --model-id dfm-sentence-encoder-medium