WordPiece

WordPiece: A subword tokenisation method building a vocabulary to represent frequent whole words and others through fragments. It is associated with BERT and reduces out-of-vocabulary problems. Segmentation depends on the learned vocabulary and need not match linguistic morphemes. This perspective connects the definition with practical behaviour in artificial intelligence systems.

In italiano: WordPieceLegge I · Embedding, feature e spazi vettoriali

It is associated with BERT and reduces out-of-vocabulary problems. Segmentation depends on the learned vocabulary and need not match linguistic morphemes.