RL-branch stands as a distributed reinforcement learning framework, which accomplishes the implementation of distributed reinforcement learning algorithms by amalgamating distributed data sampling and centralized training.

1. Components:

• Functioning as the primary program entry point, the Runner is tasked with sampling from the environment and calling upon remote Agents to execute distributed sampling and centralized training.

• The Agent embodies the intelligent algorithm, allowing users to engage in training with intelligent algorithms effortlessly, eliminating the need for coding.

2. Algorithm Support:

• The platform extends support to various algorithms such as Q-Learning, DQN, Rainbow, SAC, DDPG, TRPO, TD3, BC, GAIL and more, leveraging the intelligent algorithms furnished by the platform.
• Users can leverage these algorithms without the need for manual coding.

1. main function: Create multiple Runners and launch them automatically in a Ray cluster using the 'makeRunner' function.

   if __name__ == '__main__':
   	# Create two GymRunners using makeRunner
   	runners = makeRunner(GymRunner, num=2)
   	# run
   	run()
   

2. To Run:

• Install the package:

  pip install -e .
  

• navigate to the ddpg directory:

  cd example/ddpg
  

• start the ray cluster(head node) on port 5876:

  ray start --head --port=6379
  

• run the DDPG example script:

  python ddpg.py