Model 2 is the second iteration of undertaker optimization, with the
sidechains repacked by Rosetta.  Rosetta gave a lower energy to this
model than to the repacking of the third iteration.