Hajji-X/lixus

An NLP project for Moroccan Darija. Part of Volubilis. # Lixus A small NLP project for Moroccan Darija. Part of Volubilis. ## Status Very early. Still figuring things out, so the structure and code will keep changing. ## What this is An attempt to build a Darija model mostly from scratch, as a way to actually learn how these models work. scikit-learn is only here for quick baselines. ## Setup Needs uv and Python 3.12. ``` uv venv --python 3.12 source .venv/bin/activate uv pip install numpy scikit-learn pandas ``` ## Layout * `src/lixus/` is the package * `experiments/` holds training scripts * `data/` holds datasets, which are not committed * `notes/` is a running log ## License Apache 2.0. See LICENSE.