I went through the basics of reinforced ML in the folder of Basics
- I researched an learned about the DQN Model and I used a simple enviroment to train a model (AI) to balance a board with only 2 actions availible while giving it cart velocity, pole angle, cart position, and pole angular velocity. The cart position is only useful because the cart can fall of the track resulting in losing points. The max points earned is 500
- dqn.py is the training script.
- watch.py is a visual script using pygame that allows you to see the model (AI) use the finished weights of one of the dqn training sessions to complete the model succefully removing all random chance.