using DQN play games in OpenAI gym
Deep reinforcement learning has received more and more attention. Deep q network is the beginning of DRL. Therefore, for those who want to enter the field of deep reinforcement learning, I hope that my code can help you better understand DQN algorithm and openai-gym environment.
I use pytorch to implement the original dqn and its two variants: dueling dqn and double dqn. When doing the cartpole game, I observed these two variants, especially the dueling dqn is better than the original dqn.
most idea come from https://github.com/MorvanZhou/Reinforcement-learning-with-tensorflow