Transformers architectures have been dominating the benchmarks of most Natural Language Processing tasks for the past 3 years with new models popping up every week, making it difficult to keep up with the state-of-the-art when trying to decide which model to use for a real world scenario. The French/American startup Huggingface is helping a lot to unify the efforts to make these new models available and I will present a part of these efforts and an example demonstrating how easy it now is to use and test various models of their transformers library.