Adaptive Rational Activations to Boost Deep Reinforcement Learning | Read Paper on Bytez