Reinforcement Learning

Modular DRL framework with a focus on StarCraft II, following in DeepMind’s footsteps, replicating results. Created with performance, extensibility, and reproducibility in mind.
Works with Gym, Atari, and MuJoCo as well.