why rebo2 included in airebo?

Dear Lammps users,

What is the justification for using second generation REBO (2002)
with the AIREBO formalism (2000)?

Using AIREBO with Brenner’s original 1990 REBO1 formalism is well
justified in my opinion because REBO1 does not include torsion and
AIREBO corrects for this.

REBO2 directly includes dihedral terms in the bond order function.
Adding an external torsion term from the AIREBO formalism on top
of the REBO2 bond order dihedral term seems likely to create an
potential that is overly stiff to torsion.

Does AIREBO included in LAMMPS use parameters that have been
refit to REBO2 or is it just using the original parameters developed
for use with REBO1?

Thanks for your advice.
John Russell