The A-model partition function is a topological invariant of any hyperbolic surface. So, what is this invariant?

I'm not sure if there is a way to normalize these since the usual perscription $Z(S^2) = 1$ doesn't work here, but I figure there's a way to do it anyway.