In this thesis, a data-driven approach that utilizes transformer-based neural network architectures to automatically determine optimal weighting parameters for MPC is investigated by extending the results recently presented in the literature for linear control algorithms.