v0.4.0

Features

• New algorithm: DQN
  • Fully generic implementation of the DQN algorithm. Users can create their own RL environment and model structure, and after implementing certain traits they can easily apply the DQN algorithm.
• New gym environment: CartPole
• New example: dqn_cartpole
• Greatly improved viz
  • Added a log tab to capture and view logs in viz
  • Heatmap for plot data to have a greater sense of the density of samples
  • Help popup for keybinds

Other

• Added many unit tests
• Improved documentation
• Improved examples