Table 1

Different configurations for eight algorithms in Figures 1 and 2.

algorithmbatch norm1-layer lstm2-layer lstmca-blocksa-blockha-blockscalar weightsmhsa-blockvector-based attention weights
Figure 1a 
Figure 1b 
Figure 1c 
Figure 1d 
Figure 1e 
Figure 1f 
Figure 2a 
Figure 2b 

or Create an Account

Close subscription notice
Close access options