rl-framework

General framework of reinforcement learning algorithms. Implemented for experiments of AILab, Peking University.

Basic Rules

• Universal
• Easy to use / edit
• Documents

Algorithms

• DDPG (Deep Deterministic Policy Gradient), ICLR 2016 (https://arxiv.org/abs/1509.02971)
• DQN (Deep Q-Network), Nature 2015 (https://www.nature.com/articles/nature14236
• TRPO (Trust Region Policy Optimization), ICML 2015 (https://arxiv.org/abs/1502.05477)
• PPO (Proximal Policy Optimization), arXiv 2017 (https://arxiv.org/abs/1707.06347)
• some other algorithms...

TODOLIST

• Try to design an easy-to-use mode for lab's experimental needs
• Evaluate the PyTorch & TensorFlow implement of DDPG, and create an easy document for DDPG
• Implement basic DQN published on Nature
• READ TRPO and PPO BEFORE implement them
• Think about other algorithms need to be added in this repo
• some other todos...