That's an interesting dynamic of the training data impacting the architecture. I wonder if this is a one-off or we see that in other areas as well.

I think this is inevitable. Whatever is most highly represented (correctly) will become even more dominant.