That makes sense -- these are details that would be good to have in the methodology :)

Would be interested on the results from running a similar experiment with more powerful models, though that's more of a curiosity thing than anything lol