kevin-bretz/mario-tas-ppo

Super Mario World TAS Agent (PPO weights)

From-scratch PPO checkpoints for a group project comparing DQN, PPO and SAC on speedrunning the first level of Super Mario World (SNES), for the Modern Game AI course at Leiden University.

Code and full details: https://github.com/kevin-bretz/mario-tas-ppo

• best_model.pth is the best-performing policy and value network.
• progress.pth is the latest training checkpoint, for resuming training.