Project Structure

• src/ – contains all source code, organized into:

  • reinforcement/ – approaches based on classical reinforcement learning (e.g., PPO).
  • non_reinforcement/ – alternative methods (e.g., MDP, POMDP, Neuroevolution).

• notebooks/ – Jupyter notebooks for training, experimentation, and evaluation.

• runs/ – stores training artifacts such as models, videos, and plots from each training session.

Dependencies

This project relies on the following open-source libraries:

• gym
• gym-super-mario-bros (https://github.com/Kautenja/gym-super-mario-bros)
• nes-py
• torch
• numpy
• matplotlib

All rights for the game "Super Mario Bros" belong to Nintendo. The gameplay videos in this repository are for academic use only and were created using the gym-super-mario-bros environment.