Input Label Model

AbstractNeural text classification models typically treat output labels as categorical variables that lack description and semantics. This forces their parametrization to be dependent on the label set size, and, hence, …...