Install

PNPL requires Python 3.10+ and installs from PyPI:

pip install pnpl

Core scientific dependencies include numpy, pandas, torch, h5py, mne, mne_bids, huggingface_hub, and requests (used by the OSF and Radboud download backends).

Authentication for gated datasets

The Radboud-hosted datasets (Armeni2022, Schoffelen2019) require an approved data-sharing agreement plus credentials. Set them in your environment (or a project-local .env next to your code — pnpl loads .env automatically):

export RADBOUD_USERNAME="you@orcid.org"   # often an ORCID
export RADBOUD_PASSWORD="..."

LibriBrain (Hugging Face) and MEG-MASC (OSF) are open and require no credentials.

Development install (editable)

git clone https://github.com/neural-processing-lab/pnpl.git
cd pnpl
python -m venv .venv && source .venv/bin/activate
pip install -e .