Short Text Paraphrase Identification Model Based on RDN-MESIM

Table 1

The results in different depths of RDN’s dense block and transition layer pair.

Layer	Loss	Accuracy	Trainable parameter
2	0.0956	0.9710	1,933,570
3	0.1363	0.9615	3,257,602
4	0.1271	0.9610	4,977,922
5	0.1351	0.9638	7,094,530
6	0.1768	0.9412	9,607,426