The LSTM tagger above is typically sufficient for part-of-speech tagging, but a sequence model like the CRF is really essential for strong performance on NER.
確定! 回上一頁