2 Matching Annotations
  1. Feb 2020
    1. revious research on word embeddings hasshown that sparse representations, which canbe either learned on top of existing denseembeddings or obtained through model con-straints during training time,