They were trained pretty much on the same data.