Improving Neural Machine Translation with AMR Semantic Graphs

Table 1

Statistics on the corpus.

Corpus	Sentences	Tokens (English)	Tokens (Vietnamese)
Training	133K	2.44 M	2.87 M
Fine tuning (tst2012)	1553	28 K	34 K
Test 1 (tst2013)	1268	26.7 K	33.6 K
Test 2 (tst2015)	1080	21 K	26.2 K