NMT十篇必读论文（九）Sequence to Sequence Learning with Neural Networks

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

 

又回到神经机器翻译，这篇论文present a general end-to-end approach to sequence learning that makes minimal assumptions on the sequence structure。使用了两个LSTM作为encoder和decoder（没用attention），生词用UNK表示，在WMT14的数据集上比传统的基于短语的统计机器翻译提高了近2个bleu值。

 

论文的大体思想和RNN encoder-decoder是一样的，只是用来LSTM来实现。

paper提到三个important point：

1）encoder和decoder的LSTM是两个不同的模型

2）deep LSTM表现比shallow好，选用了4层的LSTM

We found that the LSTM models are fairly easy to train. We used deep LSTMs with 4 layers,with 1000 cells at each layer and 1000 dimensional word embed