Comment by vcryan

For the benchmark... it seems a bit silly to compare only to outdated/underperforming models.