Table 2 State/action dimensionality of the additional scalability benchmarks (PyBullet + Isaac).

From: Risk sensitive twin distributional critics with a lambda lower confidence bound for continuous control reinforcement learning