Repository contains Jupyter Notebooks for training reinforcement learning agents on Atari games using Gymnasium environments. It focuses on training Lunar Lander with A2C and PPO algorithms, demonstrating PPO's stable performance over 300,000 steps.
From the README
Training Lunar Lander-Atari game for 300,000 steps between A2C and PPO algorithms validating PPO has stable performance .
Skills
- Jupyter Notebook
- atari-games
- gymnasium
- reinforcement-learning
- tensorboard