An implementation of Proximal Policy Optimization (PPO) using Generalized Advantage Estimation (GAE) and multi-processing.
python3 -m venv venv to set up a virtual environment
cd pip install . to install, or pip install -e . for development.
python src/run_ppo.py config/pendulum.yaml trains PPO for a given config file. Examples for different environments with hyperparameters I've found that work well can be found in config/.
See https://salmanmohammadi.github.io/content/ppo/ for an explanation of the method.