DQN

Reproduce DQN with PARL

Based on PARL, we provide a simple example of the DQN algorithm.

• Paper: DQN in Human-level Control Through Deep Reinforcement Learning

Result

Performance of DQN playing CartPole-v0

How to use

Dependencies:

• paddlepaddle>=2.0.0
• parl>=2.0.0
• gym

Start Training:

# To train a model for CartPole-v0 game
python train.py

DQN-Variants

For DQN variants such as Double DQN and Dueling DQN, please check here