SimBa: Simplicity Bias for Scaling Parameters in Deep Reinforcement Learning

sonyresearch.github.io