No, but the tokens would be similar and reinforced by the model, so understanding isnt pertinent in this case