https://github.com/lebellig/shut-the-box-9
https://github.com/lebellig/shut-the-box-9
Last synced: about 1 month ago
JSON representation
- Host: GitHub
- URL: https://github.com/lebellig/shut-the-box-9
- Owner: lebellig
- Created: 2022-07-28T23:34:56.000Z (almost 4 years ago)
- Default Branch: master
- Last Pushed: 2022-08-10T15:26:22.000Z (almost 4 years ago)
- Last Synced: 2025-09-08T12:48:02.624Z (11 months ago)
- Language: Python
- Size: 29.3 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Shut the box 9
This project aims at using reinforcement learning techniques to master the game named "Shut the box 9". ([Rules](https://en.wikipedia.org/wiki/Shut_the_box)).
The idea here is to consider this game as a great exercice to implement reinforcement learning algorithms and to learn to train models with the _Jax_ framework (with the _Flax_ library for the neural networks).
## Project structure
We only propose algorithms that do not rely on humain annotated games or humain played games. All the data pipeline can be found in the `/src/data` folder.
The game in itself is implemented in the `/src/game` folder and a memory buffer of the played moves is added to the scripts.
The trained (or random) agents are available in `/src/agents`. All the scripts dealing with agents training and neural networks architectures are in `/src/learn`.
Some tests are also implemented to ensure the robustness of the training pipeline in the `/tests` folder
## Installation
1. Clone this repository:
```bash
git clone git@github.com:gle-bellier/shut-the-box-9.git
```
2. Install requirements:
```bash
cd shut-the-box-9
pip install -r requirements.txt
pip install -e .
```