This project trains embeddings for scientific articles using data collected from the INSPIRE-HEP API.

The embeddings are used by the hep-recommender web application to provide similar article recommendations in the field of High Energy Physics.

To test the package and train the embeddings, start a virtual environment and install the necessary dependencies using

pip install -r requirements.txt

Place the file 'test_references_data.txt' in the 'input_data' folder. Then run

python model_training.py

This will save the trained embeddings on the 'output_data' directory. Adjust the 'model_config.json' if desired.