Short-Term Load Forecasting (STLF) is the electrical demand prediction over short horizons. Nowadays, power systems must deal with several challenging aspects: the increase in efficiency of end-uses, the growing share of electricity in end-uses, and the increasing relevance of distributed energy resources. To match demand to generation at the economic and energy efficiency optimum, the risen share of variable renewable energy in power mixes and the need for efficient load management strategies call for increasingly accurate load predictions. Conventional approaches to STLF used univariate data and statistical models, prioritizing interpretability and computational inexpensiveness over predictive accuracy. Progresses in deep learning (DL) algorithms have made multivariate data and DL models the state-of-the-art approach.