Weightless Neural Networks library (wnnlib)

Setup

Clone the repository:

git clone https://github.com/stivenschwanz/wnnlib
cd wnnlib

Create the virtual environment:

pip3 install virtualenv
python3 -m venv .venv

Add the source folder permanently to the Python path:

echo "$(pwd)/src" > `echo $VIRTUAL_ENV/lib/python*/site-packages/`src.pth

Activate the virtual environment:

source .venv/bin/activate

Install required packages:

pip3 install -r requirements.txt

Run experiments

Activate the virtual environment:

source .venv/bin/activate

Toy anomaly detection problems:

python3 -m unittest src/wnnlib/algos/NPCLAD.py

Run unit tests

Test VGRAM node:

python3 -m unittest src/wnnlib/vgram/VGRAMNode.py

Test VGRAM array:

python3 -m unittest src/wnnlib/vgram/VGRAMArray.py

Test fixed scalar codec:

python3 -m unittest src/wnnlib/codecs/FixedScalarCodec.py

Test adaptive scalar codec:

python3 -m unittest src/wnnlib/codecs/AdaptiveScalarCodec.py

Test flex scalar codec:

python3 -m unittest src/wnnlib/codecs/FlexScalarCodec.py

Test KD-tree vector codec:

python3 -m unittest src/wnnlib/codecs/KDTree.py

Test binary utilities:

python3 -m unittest src/wnnlib/utils/BitUtils.py

Generate distribution archives

Make sure you have the latest version of PyPA’s build installed:

python3 -m pip install --upgrade build

Build the distribution archives:

python3 -m build

Install twine to upload the distribution packages:

python3 -m pip install --upgrade twine

Upload the distribution packages:

python3 -m twine upload --repository testpypi dist/*