3 ms·Fast reinforcement learning through composition of behavior29 points by atg_abhishek 6y agoilaksh 6y agoSeems like the world model representation and accuracy is more important than the reinforcement part.