Effects of sparse rewards of different magnitudes in the speed of learning of model-based actor critic methods | Read Paper on Bytez