Hierarchical Self-Attention Hybrid Sparse Networks for Document Classification

Table 5

Results in classification accuracy.

Models	Yelp 2018	IMDB
HAHNN^∗ [2]	73.28	95.17
byte mLSTM7 [12]	—	92.2
Bigbird^∗ [15]	72.16	95.2
BERT-ITPT-FIT [24]	70.58	—
LSTM-reg [25]	68.7	—
MPAD-path^∗ [14]	—	91.84
HMAN^∗ [26]	73.4
H-CRAN^∗ [27]	73.0
CapsNet [28]		89.72
Ours	73.48	95.69