bmazoure/ppo_jax

Jax implementation of Proximal Policy Optimization (PPO) specifically tuned for Procgen, with benchmarked results and saved model weights on all environments.

24
/ 100
Experimental

This project offers a JAX-based implementation of the Proximal Policy Optimization (PPO) algorithm, specifically tailored for reinforcement learning tasks within the OpenAI Procgen environments. It takes in game environment definitions and outputs trained PPO policies, along with pre-trained weights for various Procgen games. It's designed for machine learning researchers and practitioners focused on developing and evaluating agents for procedurally generated game environments.

No commits in the last 6 months.

Use this if you are a researcher or practitioner working with reinforcement learning and need a stable, benchmarked JAX implementation of PPO for OpenAI's Procgen game environments.

Not ideal if you are looking for a general-purpose PPO implementation for non-Procgen environments or if you are not familiar with JAX, Flax, and Optax.

reinforcement-learning game-AI PPO deep-learning-research procedural-generation
No License Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 8 / 25
Maturity 8 / 25
Community 8 / 25

How are scores calculated?

Stars

59

Forks

4

Language

Python

License

Last pushed

Aug 04, 2022

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/bmazoure/ppo_jax"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.