Mentioned as a baseline for comparison, their model (1.6) was outperformed by Size Zero.
Stanford Online