The product learns by using a piece of text from the data (say, the opening sentence of the Wikipedia article) and endeavoring to predict another token within the sequence. It then compares its output with the particular text while in the training corpus and adjusts its parameters to proper any erro