1 paper
Jacob A. Matthews, John R. Starr, Marten van Schijndel
Pretrained language model (PLM) hidden states are frequently employed as contextual word embeddings (CWE): high-dimensional representations that encode semantic information given l…