Reinforcement Learning library for public, for now, it only supports TensorFlow.
pip install rockrl
RL algorithms are implemented to support gymnasium==0.29.1
version. Main requirements are that:
env.reset()
would returnstate
andinfo
states.env.step(action)
would returnstate
,reward
,terminated
,truncated
,info
states.
- PPO (Discrete and Continuous)
RockRL/tensorflow/examples/ppo/LunarLander-v2/LunarLander-v2.py
is an example of using PPO to solve LunarLander-v2 (Discrete) environment.RockRL/tensorflow/examples/ppo/BipedalWalker-v3/BipedalWalker-v3.py
is an example of using PPO to solve BipedalWalker-v2 (Continuous) environment.RockRL/tensorflow/examples/ppo/BipedalWalkerHardcore-v3/BipedalWalkerHardcore-v3.py
is an example of using PPO to solve BipedalWalker-v3 Hardcore (Continuous) environment.