CISC 856 Reinforcement Learning Project Hugh Morison (10179903) Matthew Filipovich (20029031)
All the code is structured as a python package called 'olympia'. A sample training run (over 5 episodes) can be performed by running main.py. Running play_trained_agents.py will render an episode using the best performing agents. All the trained models are *.h5 files contained in agent_models/.
The code for this project was written using Python 3.6.x in both MacOS and Windows 10.
Required python packages:
- gym
- keras
- tensorflow
- numpy