Self Training Sneak Game with Reinforcement Learning, Python, Pytorch and Pygame
This Model was trained during almost 10 minutes of Gameplay, and using concepts such as Agent, Reward, Enviroment, State and Memory. At this current version, The Model can score 40 or more points.
To test the Saved Model, run agent_with_saved_model.py, or to Train a new Model, run agent_with_new_model.py
Check-it out
Sneak Game with Reinforcement Learning Preview: https://youtu.be/FiyK-_bB75g
Code: https://github.com/mendez-luisjose/Sneak-Game-with-Reinforcement-Learning