billchent / reinforcement_learning_with_Tensorflow

Minimal implementations of reinforcement learning algorithms by Tensorflow

Reinforcement Learning with Tensorflow

This is a series of tutorials about Reinforcement Learning implemented by Tensorflow.
These tutorials don't aim to reproduce the methods used in the original papers, but for algorithms in a quite easy version.
Generally they are for learning and basic understanding, but not production.

Algorithms

1. Tabular Q Learning
2. Sarsa
3. DQN
4. Double DQN
5. Double DQN 2.0
6. Double DQN with Portional Preoritized Experience Replay
7. Dueling Double DQN
8. Policy Gradient
9. Actor Critic
10. DDPG
11. A3C
12. A2C+PPO

More will be updated in the near future.

Dependency:

1.Python3
2.Tensorflow
3.gym
4.numpy
5.matplotlib

Setup and Run

git clone https://github.com/wangshuailong/reinforcement_learning_with_Tensorflow.git  
cd DQN (Or other folds you want)  
pytho3 run.py  

Reference:

1. Movan Zhou
2. Jaromír
3. relaax