Is There a Difference Between the Second Generation and the First Generation?

The ability to improve is like much more available because you didn't have to take as many steps as it gets to the same point in your previously app. The obvious questions are like, why can't we just do like entry-gradualization or like wait decay or something? And maybe those approaches would work equally well, I don't know. But then you have to do like a goal as well and compare against just training the same agent for a little bit longer. Asymptotically, the second generation agent is better than the first generation agent. Probably most of the effect is not coming from training for longer.