Computer scienceData scienceNLPLanguage representationTransformers

BERT Transformer

BERT tokenization

Report a typo

Choose the benefit BERT's subword tokenization offer over traditional word-based tokenization.

Select one option from the list

It enables BERT to process words independently

It eliminates the need for the [SEP] token

It handles rare words, out-of-vocabulary terms, and morphologically rich languages

It ensures that all sequences have the same length during training

___

Create a free account to access the full topic