Hierarchical Self-Attention Hybrid Sparse Networks for Document Classification

Table 6

Parameters comparison with variants models.

	Param	Pruning rate		Param	Pruning rate
GRU	79800	—	LSTM	106400	—
GRU1	47000	41.26%	LSTM1	57200	46.24%
GRU2	46600	41.60%	LSTM2	56600	46.80%
GRU3	27000	66.17%	LSTM3	27200	74.44%