/jrl

Primary LanguagePython

jrl : JAX for Reinforcement Learning

Goal

Implement as many RL algorithms as possible in JAX

Completed

Algorithm Action Space Method File
DDPG Continuous Model-Free ddpg.py
TD3 Continuous Model-Free td3.py
SAC - learned temperature Continuous Model-Free sac.py
DrQ Continuous Model-Free drq.py
DroQ Continuous Model-Free sac.py

Wishlist

Algorithm Action Space Method File
DQN Discrete Model-Free
Rainbow Discrete Model-Free
Planet Continuous/Discrete Model-based
Dreamer Continuous/Discrete Model-based
DreamerV2 Continuous/Discrete Model-based
TRPO Continuous/Discrete Model-based
PPO Continuous/Discrete Model-based
DrQv2 Continuous Model-free
SAC - fixed temperature Continuous Model-Free

Sources