Stable Baslines with JAX & Haiku

Implementation of Stable Baselines based on JAX & Haiku.

This library is based on Stable Baselines 3 (https://github.com/DLR-RM/stable-baselines3).

Implemented Algorithms

Name	Online_learning	Box	Discrete	MultiDiscrete	MultiBinary
BC	❌	✔️	✔️	❌	❌
OnlineBC	✔️	✔️	✔️	❌	❌
DT	❌	✔️	❌	❌	❌
DU	❌	✔️	✔️	❌	❌
SAC	✔️	✔️	✔️	❌	❌
PPO	✔️	✔️	✔️	❌	❌

Install

git clone https://github.com/kwk2696/sb3-jax-haiku.git
pip install -e .

Benchmark

We use Intel i9-10940X, RTX 3090 to benchmark Decision Transformer (DT) on MuJoCo Ant environment.

Name	sb3-torch	sb3-jax-haiku
SAC	163 step / sec	236 step / sec
DT	0.03 step / sec	3 step / sec

Example

Example codes are available in tests directory.

Currently Working On ...

TD3, Generative modeling for RL (e.g. diffuser)