Is There a Random Noise Vector?

The idea was mostly about scaling to quite simple things like for example, the masses heavier and you push. And maybe it can then adapt when the mass of the actual thing is double or half. So if your state is the current position of it and you pushing it, then the change of state was how much it moves,. Then if you scale how much it moved, then maybe that like to trade in policy in the case where it moves double the amount and the cases where it moves half the amount. The key thing was that it was dynamic augmentation rather than like the actual states because if you change just the states, that's really good for image based learning as well by doing a rad