“NLP Status Report 2016-10-17”版本间的差异

来自cslt Wiki
跳转至： 导航、 搜索
 
（2位用户的5个中间修订版本未显示）
第9行： 第9行：
 
*Read papers for reading group
 
*Read papers for reading group
 
||
 
||
−
*finished the code of RNNG
+
*finish the code of RNNG and get an initial result.
 
|-
 
|-
 
|Jiyuan Zhang ||
 
|Jiyuan Zhang ||
第18行： 第18行：
 
*solve bug that my model training process is not in accordance with qixin's model using the adadeltaOptimazer
 
*solve bug that my model training process is not in accordance with qixin's model using the adadeltaOptimazer
 
*code and run memory network
 
*code and run memory network
 +
*By Yang: 1) fix the code of poetry generation; 2) understand the paper of MemNN; 3) read through the open-source code of MemNN and discuss with me next Monday.
 
|-
 
|-
 
|Aodong Li ||  
 
|Aodong Li ||  
第29行： 第30行：
 
||
 
||
 
*understand Neural Machine Translation
 
*understand Neural Machine Translation
−
*get familiar with Theano
+
*get familiar with Tensorflow
 +
*By Yang: 1) manage to run the code of NMT with Tensorflow; 2) Read through the code and discuss with me next Week.
 
|-
 
|-
 
|Shiyao Li ||
 
|Shiyao Li ||
第37行： 第39行：
 
|-
 
|-
 
|Shiyue Zhang ||  
 
|Shiyue Zhang ||  
−
 
+
* write the RNNG Use Guide [[http://cslt.riit.tsinghua.edu.cn/mediawiki/images/2/2b/RNNG_Code_Use_Guide_simplified.pdf RNNG Use Guide]]
 +
* read the RNNG code again, and discuss it with Teacher Feng
 +
* share a paper
 
||
 
||
−
 
+
* read the code of Memory Network
 +
*By Yang: understand the code of MemNN and discus with me this week.
 
|-
 
|-
 
|}
 
|}

2016年10月17日 (一) 07:17的最后版本

Date People Last Week This Week
2016/10/17 Yang Feng
  • modified the code of MemNN to RNNG
  • discussed the code of MemNN and RNNG with Andy and Shiyue
  • modified the code of RNNG to add MemNN
  • Read papers for reading group
  • finish the code of RNNG and get an initial result.
Jiyuan Zhang
  • shared a paper
  • Solved the bug that my model cost was not in accordance with qixin's model
  • solved some bugs about my model
  • solve bug that my model training process is not in accordance with qixin's model using the adadeltaOptimazer
  • code and run memory network
  • By Yang: 1) fix the code of poetry generation; 2) understand the paper of MemNN; 3) read through the open-source code of MemNN and discuss with me next Monday.
Aodong Li
Andi Zhang
  • fully understood Memn2n torch code
  • shared a paper
  • understand Neural Machine Translation
  • get familiar with Tensorflow
  • By Yang: 1) manage to run the code of NMT with Tensorflow; 2) Read through the code and discuss with me next Week.
Shiyao Li
Shiyue Zhang
  • write the RNNG Use Guide [RNNG Use Guide]
  • read the RNNG code again, and discuss it with Teacher Feng
  • share a paper
  • read the code of Memory Network
  • By Yang: understand the code of MemNN and discus with me this week.
取自“http://cslt.org/mediawiki/index.php?title=NLP_Status_Report_2016-10-17&oldid=23097”