A Study on Chinese-English Machine Translation Based on Transfer Learning and Neural Networks

Table 5

Experimental results of the monolingual-to-machine translation quality assessment model based on self-attentive networks under joint training conditions.

Method	Data	Pearson’s	MAE	RMSE
Transformer+embedding+fine-tuning	Dev	0.428	0.1395	0.1841
	Test2016	0.4146	0.138	0.1739
	Test2017	0.4146	0.138	0.1739
Transformer+latent+fine-tuning	Dev	0.6814	0.1067	0.151
	Test2016	0.6874	0.1048	0.1425
	Test2017	0.6801	0.1024	0.1458
Transformer+wordpredicting+fine-tuning	Dev	0.7065	0.0995	0.1347
	Test2016	0.704	0.0994	0.1378
	Test2017	0.6957	0.1023	0.13658