Repo for misc RL algorithms:

Below is a gif of one evaluation episode in the Lunar Lander v2 environment using my implementation of PPO.

Maintainer notes:

• Experiments done using Python 3.11.9 and miniconda env.
• I currently have gymnasium==0.29.1.
  • In the v1.0 version, gym.vector.make will be replaced by gym.make_vec.
  • The current version is still widely supported though, as per gymnasium's docs, so I'll stick to it.

List of algos:

• Bandits.
• N-step Q-learning.
• Dyna-Q.
• PPO.
• DQN.