“NLP Status Report 2017-7-10”版本间的差异

来自cslt Wiki
跳转至： 导航、 搜索
第21行： 第21行：
 
|-
 
|-
 
|Shipan Ren ||
 
|Shipan Ren ||
−
* read and run ViVi_NMT code  
+
* run two versions of the code on small data sets (Chinese-English)  and tested these checkpoint
−
* read the API of tensorflow
+
    found version 1.0 save time about 0.03s  per step, and these two version  has  similar complexity and bleu values
−
* debugged ViVi_NMT and  upgraded code version to tensorflow1.0  
+
    found that the bleu is still good when the model is over fitting . (reason: the test set and the train set of small data set are similar in content and style)
−
* found the new version saves more time，has lower complexity and better bleu than before
+
* run two versions of the code on big data sets (Chinese-English) . OOM（Out Of Memory） error occurred when version 0.1 was trained using large data set，but version 1.0 worked
 +
    reason: improper distribution of resources by the tensorflow0.1 frame leads to exhaustion of memory resources
 +
    I had tried 4 times （just enter the same command）, and version 0.1 worked
 +
    found version 1.0 save time about 0.06s  per step, and these two version  has  similar complexity and bleu values
 +
* downloaded the wmt2014 data set ,used the English-French data set to run the code and found the translation is not good (reason:improper word segmentation)
 
||
 
||
−
* test two versions of the code on small data sets (Chinese-English) and large data sets (Chinese-English) respectively
+
* do word segmentation on wmt2014  data set
−
* test two versions of the code on WMT 2014 English-to-German parallel dataset and WMT 2014 English-French dataset respectively
+
* run two versions of the code on wmt2014  data set
−
* record experimental results
+
* record the result and do analysis
−
* read paper and try to make the bleu become a little better
+
* learn and train moses(use big data sets (Chinese-English))
 
|-
 
|-
 
      
 
      
  
 
|}
 
|}

2017年7月10日 (一) 05:58的版本

Date People Last Week This Week
2017/7/3 Jiyuan Zhang
  • reproduced the couplet model using moses
  • continue to modify the couplet
Aodong LI
  • Tried a seq2seq with style code model but it didn't work.
  • Coded attention-based seq2seq NMT in shallow fusion with a language model.
  • Complete coding and have a try.
  • Find more monolingual corpus and upgrade the model.
Shiyue Zhang
Shipan Ren
  • run two versions of the code on small data sets (Chinese-English) and tested these checkpoint 
    found version 1.0 save time about 0.03s  per step, and these two version  has  similar complexity and bleu values 
    found that the bleu is still good when the model is over fitting . (reason: the test set and the train set of small data set are similar in content and style)
  • run two versions of the code on big data sets (Chinese-English) . OOM（Out Of Memory） error occurred when version 0.1 was trained using large data set，but version 1.0 worked 
    reason: improper distribution of resources by the tensorflow0.1 frame leads to exhaustion of memory resources 
    I had tried 4 times （just enter the same command）, and version 0.1 worked 
    found version 1.0 save time about 0.06s  per step, and these two version  has  similar complexity and bleu values
  • downloaded the wmt2014 data set ,used the English-French data set to run the code and found the translation is not good (reason:improper word segmentation)
  • do word segmentation on wmt2014 data set
  • run two versions of the code on wmt2014 data set
  • record the result and do analysis
  • learn and train moses(use big data sets (Chinese-English))
取自“http://cslt.org/mediawiki/index.php?title=NLP_Status_Report_2017-7-10&oldid=27933”