Alkaid

Trying to implement some reinforcement learning algorithms with easy to read code, high speed and well performance.

 

Features

The supported algorithms currently include:

• DQN (Deep Q-Networks)
  • DQN
  • Double DQN
  • Dueling DQN

Other features:

• Modularity and Readability: Implement agents, networks, replay buffers, etc as configurable modules and in a easy to read way
• Logging and Visualization: Easily logging training statistics and visualizing them using Tensorboard or alkaid.utils.Ploter

 

Installation

git clone https://github.com/Renovamen/alkaid.git
cd alkaid
python setup.py install

or

pip install git+https://github.com/Renovamen/alkaid.git --upgrade

 

License

Alkaid is MIT licensed, see the LICENSE file for more details.