Environments used in the RL-course at Uppsala University. Implementing some examples from "Reinforcement Learning - An introduction" by Sutton and Barto, and some other very simple environments for educational purposes.
Requirements: gymnasium and numpy
git clone https://github.com/magni84/gym_RLcourse.git
cd gym_RLcourse
pip install -e .
The code for GridWorldEnv
is based on the FrozenLake environment licensed under an MIT license.
MultiArmedBandits-v0
- Implements the multi-armed bandits of Section 2.3 in Sutton and Barto.GridWorld-v0
- Implements Example 4.1 in Sutton and Barto.GridWorld-AB-v0
- Implements Example 3.5 in Sutton and Barto.GridWorld-Windy-v0
- Implements Example 6.5 in Sutton and Barto.GridWorld-WindyKing-v0
- Same asGridWorld-Windy-v0
but with king moves.GridWorld-Maze-v0
- Implement mazes from Chapter 8 in Sutton and Barto.