Variational inference has emerged as a powerful technique for facilitating the training of intractable models, particularly in the domain of modern latent variable models. This approach addresses the challenge of computing posterior distributions, which are often intractable due to the complexity of the models involved. Variational inference transforms the problem into an optimization task, making