The aim of this project is to convert the weights of a NumPy model (.npz) into the most optimised string possible for Codingame.

pip install -r requirements.txt

Either ensure that you have a model.npz file in the current path, or change the path in the main file.

python optimizer.py

Copy the text in model.txt to the ENCODED tag of inference.py.
Modify the inference code with your model and game loop as you like.

If you find Codingame-weights-optimizer is useful, please consider giving us a star 🌟 and citing it.

@software{Codingame-weights-optimizer,
    author = {Colin de Seroux},
    month = apr,
    title = {Codingame-weights-optimizer},
    url = {https://github.com/colindeseroux/codingame-weights-optimizer},
    version = {1.0.0},
    year = {2025}
}

This project is licensed under the Apache License 2.0.
See the LICENSE file for more details.