1: \begin{abstract} 

2:  This paper investigates Frequentist consistency properties of the posterior distributions constructed via Generalized Variational Inference (\GVI) \cite{GVI}.

3:  %

4:  A number of generic and novel strategies are given for proving consistency, relying on the theory of $\Gamma$-convergence. 

5:  %

6:  Specifically, this paper shows that under minimal regularity conditions, the sequence of \GVI posteriors is strongly consistent. 

7:  %

8:  In other words, \GVI posteriors collapse to a point mass at the population-optimal parameter value $\*\theta^{\ast}$ as the number of observations goes to infinity.

9:  %

10:  These results extend to the latent variable case without additional assumptions.

11:  %s

12:  Lastly, the paper explains how to apply the results to a selection of \GVI posteriors with especially popular variational families. For example, consistency is established for \GVI methods using the mean field normal variational family, normal mixtures, Gaussian process variational families  as well as neural networks indexing a normal (mixture) distribution.

13:  %Lastly, some applications of the results are provided

14:  %

15:  %Lastly, the paper delivers a proof for the asymptotic normality of \GVI based on R\'enyi's $\alpha$-divergence under the same regularity conditions as for standard Variational Inference (\VI) as established recently \citep{VBAsymptoticNormality}. 

16: \end{abstract}