Experiments with GPT-3-based difficulty estimation

The dataset used for the experiments, containing three datasets: Literature de jeunesse libre (LjL), a collection of sentences collected from the Internet (sentencesInternet), and a collection of sentences from literature books (sentencesBooks).