“NLP Status Report 2016-12-26”版本间的差异

来自cslt Wiki
跳转至： 导航、 搜索
（以“{| class="wikitable" !Date !! People !! Last Week !! This Week |- | rowspan="6"|2016/12/19 |Yang Feng || *s2smn: wrote the manual of s2s with tensorflow http:/...”为内容创建页面）
 
 
（6位用户的10个中间修订版本未显示）
第2行： 第2行：
 
!Date !! People !! Last Week !! This Week
 
!Date !! People !! Last Week !! This Week
 
|-
 
|-
−
| rowspan="6"|2016/12/19
+
| rowspan="6"|2016/12/26
 
|Yang Feng ||
 
|Yang Feng ||
−
*[[s2smn:]] wrote the manual of s2s with tensorflow [[http://cslt.riit.tsinghua.edu.cn/mediawiki/images/5/51/Nmt-tensorflow-mannua-yfeng.pdf nmt-manual]]
+
*[[s2smn:]] read six papers to fix the details of our model;
−
*wrote part of the code of mn.
+
*wrote the proposal of lexical memory and discussed the details with Teach Wang;
−
*wrote the manual of Moses [[http://cslt.riit.tsinghua.edu.cn/mediawiki/images/9/92/Moses%E6%93%8D%E4%BD%9C%E6%89%8B%E5%86%8C--%E5%86%AF%E6%B4%8B.pdf moses-manual]]
+
*finished coding of only adding attention to the decoder and under debugging;
−
*[[Huilan:]] fixed the problem of syntax-based translation.
+
*refine Moses manual [[http://cslt.riit.tsinghua.edu.cn/mediawiki/images/9/92/Moses%E6%93%8D%E4%BD%9C%E6%89%8B%E5%86%8C--%E5%86%AF%E6%B4%8B.pdf manual]] ;
−
*sort out the system and corresponding documents.
+
*prepare the dictionary for the memory loading;
 +
*[[Huilan:]] documentation
 
||
 
||
−
*[[s2smn:]] finish the code of adding mn.
+
*[[s2smn:]] run the ecperiments.
−
*[[Huilan:]] handover.
+
*[[rnng+mn:]] try to find the problem.
 
|-
 
|-
 
|Jiyuan Zhang ||
 
|Jiyuan Zhang ||
−
*coded tone_model,but had some trouble
+
*integrated tone_model to attention_model for insteading manul rule,but the effect wasn't good
−
*run global_attention_model that decodes four sentences, [http://cslt.riit.tsinghua.edu.cn/mediawiki/images/d/d5/Four_local_atten.pdf four][http://cslt.riit.tsinghua.edu.cn/mediawiki/images/0/05/Five_local_attention.pdf five]generated by local_attention model
+
*replacing　all_pz rule with half_pz
 +
*token a classical Chinese as input,generated poem [http://cslt.riit.tsinghua.edu.cn/mediawiki/images/3/33/Story_input.pdf]
 
||  
 
||  
 
*improve poem model   
 
*improve poem model   
第23行： 第25行：
 
*coded a script for bleu scoring, which tests the five checkpoints auto created by training process and save the one with best performance
 
*coded a script for bleu scoring, which tests the five checkpoints auto created by training process and save the one with best performance
 
||
 
||
−
*extract encoder outputs
+
*
 
|-
 
|-
 
|Shiyue Zhang ||  
 
|Shiyue Zhang ||  
第29行： 第31行：
 
* tried different scale vectors, and found setting >=-5000 is good
 
* tried different scale vectors, and found setting >=-5000 is good
 
* tried to change cos to only inner product, and inner product is better than cos.
 
* tried to change cos to only inner product, and inner product is better than cos.
−
* [[http://cslt.riit.tsinghua.edu.cn/mediawiki/images/9/9f/RNNG%2Bmm_experiment_report.pdf]]
+
* [[http://cslt.riit.tsinghua.edu.cn/mediawiki/images/9/9f/RNNG%2Bmm_experiment_report.pdf report]]
−
* read a paper [[http://cslt.riit.tsinghua.edu.cn/mediawiki/images/9/92/DEEP_BIAFFINE_ATTENTION_FOR_NEURAL_DEPENDENCY_PARSING.pdf]]
+
* read 3 papers [[http://cslt.riit.tsinghua.edu.cn/mediawiki/images/9/92/DEEP_BIAFFINE_ATTENTION_FOR_NEURAL_DEPENDENCY_PARSING.pdf]] [[http://cslt.riit.tsinghua.edu.cn/mediawiki/images/a/aa/Simple_and_Accurate_Dependency_Parsing_Using_Bidirectional_LSTM_Feature_Representations.pdf]] [[http://cslt.riit.tsinghua.edu.cn/mediawiki/images/f/fb/Bi-directional_Attention_with_Agreement_for_Dependency_Parsing.pdf]]
 
* trying the joint training, which got a problem of optimization.  
 
* trying the joint training, which got a problem of optimization.  
 
||
 
||
−
 
+
* try the joint training
 +
* read more papers and write a summary
 
|-
 
|-
 
|Guli ||
 
|Guli ||
−
*read papers about Transfer learning and solving OOV
+
* finished the first draft of the survey
−
*conducted comparative test
+
* voice tagging 
−
*writing survey
+
 
||
 
||
−
* complete the first draft of the survey 
+
* morpheme-based nmt
 +
* improve nmt with monolingual data
 
|-
 
|-
 
|Peilun Xiao ||
 
|Peilun Xiao ||
−
*use LDA to generate 10-500 dimension document vector in the rest datasets
+
*learned tf-idf algorithm
−
*write a python code about a new algorithm about tf-idf
+
*coded tf-idf alogrithm in python,but found it not worked well
 +
*tried to use small dataset to test the program
 
||
 
||
−
*debug the code
+
*use sklearn tfidf to test the dataset
 
|}
 
|}

2016年12月26日 (一) 05:16的最后版本

Date People Last Week This Week
2016/12/26 Yang Feng
  • s2smn: read six papers to fix the details of our model;
  • wrote the proposal of lexical memory and discussed the details with Teach Wang;
  • finished coding of only adding attention to the decoder and under debugging;
  • refine Moses manual [manual] ;
  • prepare the dictionary for the memory loading;
  • Huilan: documentation
Jiyuan Zhang
  • integrated tone_model to attention_model for insteading manul rule,but the effect wasn't good
  • replacing　all_pz rule with half_pz
  • token a classical Chinese as input,generated poem [1]
  • improve poem model
Andi Zhang
  • coded to output encoder outputs and correspoding source & target sentences(ids in dictionaries)
  • coded a script for bleu scoring, which tests the five checkpoints auto created by training process and save the one with best performance
Shiyue Zhang
  • tried to add true action info when training gate, which got better results than no true actions, but still not very good.
  • tried different scale vectors, and found setting >=-5000 is good
  • tried to change cos to only inner product, and inner product is better than cos.
  • [report]
  • read 3 papers [[2]] [[3]] [[4]]
  • trying the joint training, which got a problem of optimization.
  • try the joint training
  • read more papers and write a summary
Guli
  • finished the first draft of the survey
  • voice tagging
  • morpheme-based nmt
  • improve nmt with monolingual data
Peilun Xiao
  • learned tf-idf algorithm
  • coded tf-idf alogrithm in python,but found it not worked well
  • tried to use small dataset to test the program
  • use sklearn tfidf to test the dataset
取自“http://index.cslt.org/mediawiki/index.php?title=NLP_Status_Report_2016-12-26&oldid=25059”