“NLP Status Report 2017-9-4”版本间的差异

• code refactoring
• wrote a document[1]

• checked experimental results and found some small problems，like follows：

   sentence before tokenize：
   "If no more people smoked, this rate would increase to at least 50%," says André Beaulieu, 
   spokesman for the Canadian Cancer Society.
   sentence after tokenize：
   " if no more people smoked , this rate would increase to at least 50 % , " says andré beaulieu , 
   spokesman for the canadian cancer society .
   It was just a equivalent replacement

• wrote the first draft paper of ViVi_NMT [2]
• pushed the baseline code,readme file and user manual to cslt github(https://github.com/CSLT-THU/CSLT_NMT)

• read some translating phrases papers
• learn more about the memory mechanism

• Clean up the code of predict.py

• configuration environment of TF-0.12 or update the code to adapt TF-1.0