Replacing your Word Embeddings by Contextualized Word Vectors

Influenced from Mikolov et al. (2013) and Pennington et al. (2014), word embeddings become the basic step of initializing NLP project. After that, lots of embeddings are introduced such as lda2vec (Moody Christopher, 2016), character embeddings, doc2vec and so on. Today, we have new embeddings which is contextualized word embeddings. The idea is similar but they achieve the same goal…