Andre Viera: That's a really interesting result. And the power of priors, right? Like how many times do we try like throwing our models at something from scratch, treat it as like a narrow problem in that sense, and train just for that one task. We find we don't get as good performance as a model that has a little bit, you know, kind of a bit of a leg up," he says. "I think Jeremy, you can tell us more about this."