Hierarchical Self-Attention Hybrid Sparse Networks for Document Classification

Table 4

Results in variants models.

Architecture	Models	Yelp 2018	IMDB
Baseline	HSAHN + GRU	73.48	91.75
	HSAHN + LSTM	73.23	88.21
Sparse GRUs	HSAHSN + GRU1	73.22	93.63
	HSAHSN + GRU2	73.28	93.12
	HSAHSN + GRU3	73.22	95.69
Sparse LSTMs	HSAHSN + LSTM1	73.00	92.18
	HSAHSN + LSTM2	72.75	91.23
	HSAHSN + LSTM3	72.94	91.25