/RL-PPO-GAMES

Primary LanguageJupyter Notebook

1. Mario-Reinforcement-Learning

The agent was able to defeat the level 1-1 of mario. Since the training was done in level 1-1 instead of learning the mechanics of the game instead it learnt the physical map details of level 1-1. This is a good area to stop. Next time random training cycles should be done while employing the game.

Mario.webm

More details at the readme file in Mario-Reinforcement-Learning folder

2. Pokemon Reinforcement Learning

It is able to play pokemon quiet effectively and goes on to defeat brock on its own. Due to the rl-training nature it gets stuck in 1 area of the game and is not able to move forward. Next time a better exploration reward should be give.

pokemon.mp4