wangqiangneu / MT-PaperReading

Record my paper reading about Machine Translation and other related works.
36 stars 2 forks source link

19-EMNLP-Jointly Learning to Align and Translate with Transformer Models #44

Open wangqiangneu opened 4 years ago

wangqiangneu commented 4 years ago

简介

在Transformer中引入word alignment的任务,目标不是提高翻译性能,而是在翻译性能不掉的情况下提高alignment accuracy。做法很简单,翻译任务decoder依然mask future,但是对齐任务不能还mask future (因为它并不知道当前词自身是谁,因为还没预测出来呢),所以他就不用mask再走一遍decoder。

论文信息

总结