Scalable Learning of Item Response Theory Models