VizSeq uses tags to group examples of different types, for example:Language IDs in multilingual machine translationValidation / test subsetsSentence length bucket IDs