Constructing AdS flux vacua requires a variety of tools to find the appropriate flux data and validate the resulting vacuum. Among these, it is important to be able to enumerate non-perturbative corrections, which require the computation of Gopakumar-Vafa (GV) invariants. Our recent AdS construction relies on a racetrack formed by exponentially-suppressed contributions to the superpotential, where GV invariants play a key role. Furthermore, as part of our validation, we check the convergence of worldsheet instantons corrections to the Kahler potential. This is a difficult task to perform, as it requires the computation of GV invariants for models with a large number of moduli. In this talk I will discuss how we achieved this and justify why our construction is under good control.