Description: "Our model also incorporates an attention
mechanism indicating the focus allocated to
the left context words, the right context words
or the word’s characters, hence making the
prediction more interpretable"