A body of writing, documents, or text used to train a supervised learning model. The text is tokenized or vectorized before being inputted into the model.