I've researched online and seen many papers on the use of RNNs (like LSTMs or GRUs) to autocomplete for, say, a search engine, character by character. Which makes sense since it inherently predicts character-by-character in a sequential manner.

Would it be possible to use the transformer architecture instead to do search autocomplete? If so, how might such a model be adapted?



You must log in to answer this question.