pywmlg

Python package template for machine learning research

Getting started

You can use this template for your own project:

On GitHub, click the "New Repository" buttom.
Under "Repository template" select "warwick-machine-learning-group/pywmlg".
Create the repository, then clone it to your computer.
On your computer, create a new branch.
Use an IDE such as VSCode to find occurences of "pywmlg", and change them to be the name of your new repo/package.
Change the name of the directory pywmlg to be the name of your python package.
Create a pull request, then merge into the main branch once the tests have passed.

We recommend always using a virtual environment for every new project. For example, here is how to create and activate a new conda environment:

conda create -n pywmlg python=3.9
conda activate pywmlg

Install the package with pip

pip install -e .

To install pytest and linting libraries:

pip install -r requirements.txt

Run the unit tests with pytest:

pytest

docker build -t pywmlg:latest .

Keeps the code neat and tidy. We use pylint and mypy. You can change pylint settings in the .pylintrc file.

pylint pywmlg/*
mypy pywmlg

Some of the smaller linting issues can be cured by running the black formatter:

black */