NMT十篇必读论文（一）attention is all you need

清华大学NLP整理的神经机器翻译reading list中提到了十篇必读论文

https://github.com/THUNLP-MT/MT-Reading-List

本文抛弃了惯用的以CNN、RNN作为位置编码的方法，单纯依靠注意力机制以及简单的三角函数进行了位置编码，起到了不错的效果。对应模型为Tensor2Tensor框架下的Transformer模型。

GitHub地址：  https://github.com/tensorflow/tensor2tensor

解释的比较好的博客：

https://blog.csdn.net/c9Yv2cf9I06K2A9E/article/details/79023069

https://ask.hellobi.com/blog/wenwen/18695

https://blog.csdn.net/qq_41058526/article/details/80783925

https://www.jianshu.com/p/3f2d4bc126e6