Once we have the sentence converted to context windows i.e. a matrix of indexes, we have to associate these indexes to the embeddings (real-valued vector associated to each word).

Other mechanisms are necessaryfor handling these kinds of transformations.