Text model

DeBERTa

DeBERTa is a transformer based language model that improves on BERT and RoBERTa using disentangled attention, which represents each word with separate vectors for content and position, plus an enhanced mask decoder for pretraining.

Type
Text