First of all, the reference you quote uses screened exchange, so it's not possible to compare the results to LDA/GGA directly. Second, without knowing more about the details of your calculation, I can't say much. The bulk MoS2 in the database is not a bilayer unless you also add vacuum, but perhaps you did. However, even so, minimizing the forces on the atoms (by using an OptimizeGeometry block, yes) will be crucial to getting good results (even for bulk, as I noted above). Whether they agree with screened exchange I don't know, however.