zlacker

Yes, there is large amounts of public domain text available, but does anyone believe this is a restriction that was imposed when feeding the models?