Created by: n00blet

Hello, I was trying to run machine translation, and was looking at the documentation and the example to run a basic NMT model. And in your documentation, the link to download the subset of dataset is not available. Link: http://paddlepaddle.bj.bcebos.com/demo/wmt_shrinked_data/wmt14.tgz.

And in the documentation, how do I create source and target dictionary for my files. Do I have to use a different library, and what will be the format of dictionary files.

Also, is there a way to preprocess without having to merge my source and target files every time. For example, OpenNMT-PY has its own pre-process. It would be nice if I could run the sample machine translation using Paddle.