See: https://openai.com/blog/requests-for-research-2/
Solving the classic game of snake with OpenAI's PPO baseline.
Clone the git repository and use the following pip command in the downloaded directory.
pip3 install -e .
Use train.py with the following to train for a million frames. This also saves the model at the end of training.
python3 train.py
Likewise, use view.py with the following to save an mp4 file of the agent playing.
python3 view.py