Markov Population Models are a widespread formalism used to model the\ndynamics of complex systems, with applications in Systems Biology and many\nother fields. The associated Markov stochastic process in continuous time is\noften analyzed by simulation, which can be costly for large or stiff systems,\nparticularly when a massive number of simulations has to be performed (e.g. in\na multi-scale model). A strategy to reduce computational load is to abstract\nthe population model, replacing it with a simpler stochastic model, faster to\nsimulate. Here we pursue this idea, building on previous works and constructing\na generator capable of producing stochastic trajectories in continuous space\nand discrete time. This generator is learned automatically from simulations of\nthe original model in a Generative Adversarial setting. Compared to previous\nworks, which rely on deep neural networks and Dirichlet processes, we explore\nthe use of state of the art generative models, which are flexible enough to\nlearn a full trajectory rather than a single transition kernel.\n