Does anyone know if there's an open source implementation (in any language) of the HexQ algorithm for hierarchy discovery in reinforcement learning, or something like it? I'd like to evaluate it in different domains but I'm having trouble understanding how to implement it from the paper's description.