sven1977 / shine

[s]erver-[h]osted [i]ntelligent [n]eural-net [e]nvironment

shine

[s]erver-[h]osted [i]ntelligent [n]eural-net [e]nvironment

shine is a Python library for reinforcement learning (RL) algorithms with a focus on deep-neural-nets and deep-RL (as opposed to more RL-theory-focused libraries like BURLAP). shine comes with it's own RL-Game-Engine called spygame (shine + pygame) based on the PyGame library.

you should use shine if you ..

• are interested in reinforcement learning and deep-learning algorithm research
• would like to test your ideas on simple 2D game environments (MDPs), which you can create using any tiled map (tmx-file) editor (e.g. Tiled)
• would like to write 2D games in Python: You can then use the spygame library (shine's PyGame engine) even without any reinforcement learning