Does Twitter-LDA toolkit support inference for new data based on a previously trained model? Thanks, Fei