Hierarchical Self-Attention Hybrid Sparse Networks for Document Classification

Figure 3

Training loss and accuracy of variant RNNs models in Yelp dataset. (a) Training loss and accuracy of HSAHSN + GRU in Yelp. (b) Testing loss and accuracy of HSAHSN + GRU in Yelp. (c) Training loss and accuracy of HSAHSN + LSTM in Yelp. (d) Testing loss and accuracy of HSAHSN + LSTM in Yelp.

(a)

(b)

(c)

(d)