Skip to content

Latest commit

 

History

History
13 lines (8 loc) · 565 Bytes

README.md

File metadata and controls

13 lines (8 loc) · 565 Bytes

PPO

An implementation of Proximal Policy Optimization (PPO) using Generalized Advantage Estimation (GAE) and multi-processing.

Installing

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/.

Paper Notes

See https://salmanmohammadi.github.io/content/ppo/ for an explanation of the method.