"Pynguin for ML libraries" is a Master Thesis made by Lucas Berg with the aim of improving Pynguin for generating tests for machine learning libraries.

The project can be downloaded using the following command:

git clone --recurse-submodules https://github.com/BergLucas/pynguin-for-ML-libraries.git

The project requires:

• Python == 3.10.4
• pip == 23.3.1

You can setup a Python development environment by running the following commands in the project root:

python -m venv .venv
source .venv/bin/activate
pip install matplotlib==3.8.4 matplotlib-venn==0.11.10 PyLaTeX==1.4.2 scipy==1.13.0 coverage==7.5.1 -e ./pynguin

The project requires:

• Conda

You can setup a Conda development environment by running the following commands in the project root:

conda env create -f environment.yml

conda activate pynguin-for-ML-libraries

This example allows you to carry out an experiment to obtain statistics on an improvement of Pynguin.

This example requires a development environment to be set up.

You can execute the example by running the following commands:

pip install -r requirements/requirements.polars.txt

python experiment.py --modules-csv-path modules/polars.csv --results-path results/polars

This example allows you to carry out an experiment to obtain statistics on an improvement of Pynguin using Docker.

This example requires:

• Docker

You can execute the example by running the following command:

sh docker_experiment.sh requirements/requirements.polars.txt modules/polars.csv polars

All code is licensed for others under a MIT license (see LICENSE).