This is a BART-large model finetuned on roughly 58000 aligned sentence pairs in English and Middle English, collected from the works of Geoffrey Chaucer, John Wycliffe, and the Gawain Poet. <br> It includes special characters such as þ. <br> This model reflects the spelling inconsistencies characteristic of Middle English. <br> Because the model is trained largely on poetry and some prose, it is best at translating those sorts of tasks. <br> Performance can be improved by sentence tokenizing input data and translating sentence-by-sentence. <br> Removing contractions (hadn't -> had not) also boosts performance.