http://index.cslt.org/mediawiki/api.php?action=feedcontributions&feedformat=atom&user=Zhangandy 2026-04-17T11:37:43Z 用户贡献 MediaWiki 1.23.3 http://index.cslt.org/mediawiki/index.php/Schedule 2017-04-17T07:20:09Z

<p>Zhangandy：/* Daily Report */</p> <hr /> <div>=NLP Schedule=<br /> <br /> ==Members==<br /> <br /> ===Current Members===<br /> <br /> * Yang Feng (冯洋)<br /> * Jiyuan Zhang （张记袁）<br /> * Aodong Li (李傲冬)<br /> * Andi Zhang (张安迪)<br /> * Shiyue Zhang (张诗悦)<br /> * Li Gu (古丽)<br /> * Peilun Xiao (肖培伦)<br /> <br /> ===Former Members===<br /> * '''Chao Xing (邢超)''' : FreeNeb<br /> * '''Rong Liu (刘荣)''' : 优酷<br /> * '''Xiaoxi Wang (王晓曦)''' : 图灵机器人<br /> * '''Xi Ma (马习)''' : 清华大学研究生<br /> * '''Tianyi Luo (骆天一)''' ： phd candidate in University of California Santa Cruz<br /> * '''Qixin Wang (王琪鑫)''' : MA candidate in University of California<br /> * '''DongXu Zhang (张东旭)''': --<br /> * '''Yiqiao Pan (潘一桥)''' ： MA candidate in University of Sydney <br /> * '''Shiyao Li （李诗瑶）''' : BUPT<br /> * '''Aiting Liu (刘艾婷)''' : BUPT<br /> <br /> ==Work Progress==<br /> ===Daily Report===<br /> <br /> {|class=&quot;wikitable&quot;<br /> ! Date !! Person !! start!! leave !! hours ||status<br /> |-<br /> | rowspan=&quot;2&quot;|2017/04/02<br /> |Andy Zhang||9:30 ||18:30 ||8 || <br /> *preparing EMNLP<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/04/03<br /> |Andy Zhang||9:30 ||18:30 ||8 || <br /> *preparing EMNLP<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/04/04<br /> |Andy Zhang||9:30 ||18:30 ||8 || <br /> *preparing EMNLP<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/04/05<br /> |Andy Zhang||9:30 ||18:30 ||8 || <br /> *preparing EMNLP<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/04/06<br /> |Andy Zhang||9:30 ||18:30 ||8 || <br /> *preparing EMNLP<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/04/07<br /> |Andy Zhang||9:30 ||18:30 ||8 || <br /> *preparing EMNLP<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/04/08<br /> |Andy Zhang||9:30 ||18:30 ||8 || <br /> *preparing EMNLP<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/04/09<br /> |Andy Zhang||9:30 ||18:30 ||8 || <br /> *preparing EMNLP<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/04/10<br /> |Andy Zhang||9:30 ||18:30 ||8 || <br /> *preparing EMNLP<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/04/11<br /> |Andy Zhang||9:30 ||18:30 ||8 || <br /> *preparing EMNLP<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/04/12<br /> |Andy Zhang||9:30 ||18:30 ||8 || <br /> *preparing EMNLP<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/04/13<br /> |Andy Zhang||9:30 ||18:30 ||8 || <br /> *preparing EMNLP<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/04/14<br /> |Andy Zhang||9:30 ||18:30 ||8 || <br /> *preparing EMNLP<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/04/15<br /> |Andy Zhang||9:00 ||15:00 ||6 || <br /> *preparing EMNLP<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |}<br /> <br /> ===Time Off Table===<br /> <br /> {| class=&quot;wikitable&quot;<br /> ! Date !! Yang Feng !! Jiyuan Zhang <br /> |-<br /> |}<br /> <br /> ==Past progress==<br /> [[nlp-progress 2017/03]]<br /> <br /> [[nlp-progress 2017/02]]<br /> <br /> [[nlp-progress 2017/01]]<br /> <br /> [[nlp-progress 2016/12]]<br /> <br /> [[nlp-progress 2016/11]]<br /> <br /> [[nlp-progress 2016/10]]<br /> <br /> [[nlp-progress 2016/09]]<br /> <br /> [[nlp-progress 2016/08]]<br /> <br /> [[nlp-progress 2016/05/01 -- 08/16 | nlp-progress 2016/05-07]]<br /> <br /> [[nlp-progress 2016/04]]</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/NLP_Status_Report_2017-4-5 2017-04-05T02:07:52Z

<p>Zhangandy：</p> <hr /> <div>{| class=&quot;wikitable&quot;<br /> !Date !! People !! Last Week !! This Week<br /> |-<br /> | rowspan=&quot;6&quot;|2017/3/27<br /> |Yang Feng ||<br /> *tested for the baseline but cannot get the reasonable result.<br /> *debug the baseline to try to reproduce the good result but failed.<br /> *fixed the problem of nan in alpha-gamma method but the result is not good.<br /> *changed the calculation of probability for alpha-gamma method but the result is neither good.<br /> *ran Moses for cwmt zh-en translation, but the training data is case-sensitive, so need to rerun.<br /> ||<br /> *rerun Moses for cwmt zh-en and cs-en<br /> *decide to use tensorflow or theano<br /> *run experiments based on the chosen platform<br /> |-<br /> |Jiyuan Zhang ||<br /> * I did keyword expansion on the qx's model<br /> *fixed some bugs<br /> *read two papers<br /> || <br /> *improve the effect of the qx's model<br /> |-<br /> |Andi Zhang ||<br /> *revise the original oov model so that it can automatically detect oov words and translate them <br /> *deal with the situation that source word is oov but target word is not oov first<br /> *it didn't predict right<br /> ||<br /> *make the model work as what we wanted<br /> *deal with the situation that source word is oov and target word is also oov, then other situations<br /> |-<br /> |Shiyue Zhang || <br /> * got a reasonable baseline on big zhen data<br /> ||<br /> * implement mem model on this baseline, and test on big data<br /> |-<br /> |Peilun Xiao ||<br /> <br /> ||<br /> <br /> |}</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/Schedule 2017-04-05T01:45:17Z

<p>Zhangandy：/* Daily Report */</p> <hr /> <div>=NLP Schedule=<br /> <br /> ==Members==<br /> <br /> ===Current Members===<br /> <br /> * Yang Feng (冯洋)<br /> * Jiyuan Zhang （张记袁）<br /> * Aodong Li (李傲冬)<br /> * Andi Zhang (张安迪)<br /> * Shiyue Zhang (张诗悦)<br /> * Li Gu (古丽)<br /> * Peilun Xiao (肖培伦)<br /> <br /> ===Former Members===<br /> * '''Chao Xing (邢超)''' : FreeNeb<br /> * '''Rong Liu (刘荣)''' : 优酷<br /> * '''Xiaoxi Wang (王晓曦)''' : 图灵机器人<br /> * '''Xi Ma (马习)''' : 清华大学研究生<br /> * '''Tianyi Luo (骆天一)''' ： phd candidate in University of California Santa Cruz<br /> * '''Qixin Wang (王琪鑫)''' : MA candidate in University of California<br /> * '''DongXu Zhang (张东旭)''': --<br /> * '''Yiqiao Pan (潘一桥)''' ： MA candidate in University of Sydney <br /> * '''Shiyao Li （李诗瑶）''' : BUPT<br /> * '''Aiting Liu (刘艾婷)''' : BUPT<br /> <br /> ==Work Progress==<br /> ===Daily Report===<br /> <br /> {|class=&quot;wikitable&quot;<br /> ! Date !! Person !! start!! leave !! hours ||status<br /> |-<br /> | rowspan=&quot;2&quot;|2017/04/02<br /> |Andy Zhang||9:30 ||18:30 ||8 || <br /> *preparing EMNLP<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/04/03<br /> |Andy Zhang||9:30 ||18:30 ||8 || <br /> *preparing EMNLP<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/04/04<br /> |Andy Zhang||9:30 ||18:30 ||8 || <br /> *preparing EMNLP<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/04/05<br /> |Andy Zhang||9:30 ||18:30 ||8 || <br /> *<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |}<br /> <br /> ===Time Off Table===<br /> <br /> {| class=&quot;wikitable&quot;<br /> ! Date !! Yang Feng !! Jiyuan Zhang <br /> |-<br /> |}<br /> <br /> ==Past progress==<br /> [[nlp-progress 2017/03]]<br /> <br /> [[nlp-progress 2017/02]]<br /> <br /> [[nlp-progress 2017/01]]<br /> <br /> [[nlp-progress 2016/12]]<br /> <br /> [[nlp-progress 2016/11]]<br /> <br /> [[nlp-progress 2016/10]]<br /> <br /> [[nlp-progress 2016/09]]<br /> <br /> [[nlp-progress 2016/08]]<br /> <br /> [[nlp-progress 2016/05/01 -- 08/16 | nlp-progress 2016/05-07]]<br /> <br /> [[nlp-progress 2016/04]]</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/NLP_Status_Report_2017-3-27 2017-04-05T01:41:53Z

<p>Zhangandy：</p> <hr /> <div>{| class=&quot;wikitable&quot;<br /> !Date !! People !! Last Week !! This Week<br /> |-<br /> | rowspan=&quot;6&quot;|2017/3/27<br /> |Yang Feng ||<br /> *tested for the baseline but cannot get the reasonable result.<br /> *debug the baseline to try to reproduce the good result but failed.<br /> *fixed the problem of nan in alpha-gamma method but the result is not good.<br /> *changed the calculation of probability for alpha-gamma method but the result is neither good.<br /> *ran Moses for cwmt zh-en translation, but the training data is case-sensitive, so need to rerun.<br /> ||<br /> *rerun Moses for cwmt zh-en and cs-en<br /> *decide to use tensorflow or theano<br /> *run experiments based on the chosen platform<br /> |-<br /> |Jiyuan Zhang ||<br /> * I did nothing (I found my ACL mask was at borderline, so I didn't have a mind to work)<br /> || <br /> improve the effect of the qx's model<br /> |-<br /> |Andi Zhang ||<br /> *fixed the bug, turns out it rises from the unfamiliarity with numpy.resize() function<br /> *the demo model can deal with oov problem(both source word and target word are oov)<br /> ||<br /> *some paper work about graduation design<br /> *run some experiments using theano on old data set and new zh2en from lihang<br /> |-<br /> |Shiyue Zhang || <br /> * got good results from gnmt<br /> * but haven't found the way to implement our model on gnmt<br /> * trying to modify our code to make it work on big data<br /> ||<br /> * go on trying to modify our code to make it work on big data<br /> * go on looking into gnmt code<br /> |-<br /> |Peilun Xiao ||<br /> <br /> ||<br /> <br /> |}</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/NLP_Status_Report_2017-3-20 2017-04-05T01:30:54Z

<p>Zhangandy：</p> <hr /> <div>{| class=&quot;wikitable&quot;<br /> !Date !! People !! Last Week !! This Week<br /> |-<br /> | rowspan=&quot;6&quot;|2017/3/20<br /> |Yang Feng ||<br /> *went through the code and made different attempt, and managed to produce the good result (47--&gt;50 on the small zh-en data set)<br /> *wrote the cross-entropy method for alpha-gamma method, but found it is different from the built-in method.<br /> *changed to use the build-in soft-cross-entropy method and ran experiments<br /> ||<br /> *get the result for alpha-gamma method<br /> *run experiments on the big data<br /> |-<br /> |Jiyuan Zhang ||<br /> || <br /> <br /> |-<br /> |Andi Zhang ||<br /> *tried to fix the bug that prevents the prediction of EOS symbol<br /> *the output_projection matrix is out of order after the copying vector process <br /> ||<br /> *fix the bug<br /> |-<br /> |Shiyue Zhang || <br /> * learned to use google's new seq2seq code(gnmt)<br /> * ran gnmt on en-de , small zh-en, cs-en, big zh-en<br /> ||<br /> * run gnmt on new big zh-en<br /> * try to find how to implement our model on gnmt<br /> |-<br /> |Peilun Xiao ||<br /> <br /> ||<br /> <br /> |}</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/Schedule 2017-04-02T01:47:20Z

<p>Zhangandy：/* Daily Report */</p> <hr /> <div>=NLP Schedule=<br /> <br /> ==Members==<br /> <br /> ===Current Members===<br /> <br /> * Yang Feng (冯洋)<br /> * Jiyuan Zhang （张记袁）<br /> * Aodong Li (李傲冬)<br /> * Andi Zhang (张安迪)<br /> * Shiyue Zhang (张诗悦)<br /> * Li Gu (古丽)<br /> * Peilun Xiao (肖培伦)<br /> <br /> ===Former Members===<br /> * '''Chao Xing (邢超)''' : FreeNeb<br /> * '''Rong Liu (刘荣)''' : 优酷<br /> * '''Xiaoxi Wang (王晓曦)''' : 图灵机器人<br /> * '''Xi Ma (马习)''' : 清华大学研究生<br /> * '''Tianyi Luo (骆天一)''' ： phd candidate in University of California Santa Cruz<br /> * '''Qixin Wang (王琪鑫)''' : MA candidate in University of California<br /> * '''DongXu Zhang (张东旭)''': --<br /> * '''Yiqiao Pan (潘一桥)''' ： MA candidate in University of Sydney <br /> * '''Shiyao Li （李诗瑶）''' : BUPT<br /> * '''Aiting Liu (刘艾婷)''' : BUPT<br /> <br /> ==Work Progress==<br /> ===Daily Report===<br /> <br /> {|class=&quot;wikitable&quot;<br /> ! Date !! Person !! start!! leave !! hours ||status<br /> |-<br /> | rowspan=&quot;2&quot;|2017/04/01<br /> |Andy Zhang||9:30 || || || <br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |}<br /> <br /> ===Time Off Table===<br /> <br /> {| class=&quot;wikitable&quot;<br /> ! Date !! Yang Feng !! Jiyuan Zhang <br /> |-<br /> |}<br /> <br /> ==Past progress==<br /> [[nlp-progress 2017/03]]<br /> <br /> [[nlp-progress 2017/02]]<br /> <br /> [[nlp-progress 2017/01]]<br /> <br /> [[nlp-progress 2016/12]]<br /> <br /> [[nlp-progress 2016/11]]<br /> <br /> [[nlp-progress 2016/10]]<br /> <br /> [[nlp-progress 2016/09]]<br /> <br /> [[nlp-progress 2016/08]]<br /> <br /> [[nlp-progress 2016/05/01 -- 08/16 | nlp-progress 2016/05-07]]<br /> <br /> [[nlp-progress 2016/04]]</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/Nlp-progress_2017/03 2017-04-02T01:46:10Z

<p>Zhangandy：以“===Daily Report=== {|class=&quot;wikitable&quot; ! Date !! Person !! start!! leave !! hours ||status |- | rowspan=&quot;2&quot;|2017/3/01 |Andy Zhang|| 9:30||18:30 ||8|| *tested the...”为内容创建页面</p> <hr /> <div>===Daily Report===<br /> <br /> {|class=&quot;wikitable&quot;<br /> ! Date !! Person !! start!! leave !! hours ||status<br /> |-<br /> | rowspan=&quot;2&quot;|2017/3/01<br /> |Andy Zhang|| 9:30||18:30 ||8|| <br /> *tested the effect of masks on baseline_beam, turns out it is a little better than without masks<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/3/02<br /> |Andy Zhang|| 9:30||18:30 ||8|| <br /> *try to use the 'sequence_length' instead of masks<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/3/03<br /> |Andy Zhang|| 9:30||18:30 ||8|| <br /> *finished coding, needed to be further tested<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/3/06<br /> |Andy Zhang|| 9:30||18:30 ||8|| <br /> *abandoned the thought from last week<br /> *talked about graduation design with Pro. Wang<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/3/07<br /> |Andy Zhang|| 9:30||18:30 ||8|| <br /> *try to use memory to solve oov problem<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/3/08<br /> |Andy Zhang|| 9:30||18:30 ||8|| <br /> *try to use memory to solve oov problem, first try word that is not in source dict but in target dict<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/3/09<br /> |Andy Zhang|| 9:30||18:30 ||8|| <br /> *it worked<br /> *try word that is in neither source dict nor target dict<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/3/10<br /> |Andy Zhang|| 9:30||18:30 ||8|| <br /> *met with bug that the model stopped predicting &lt;EOS&gt; symbol<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/3/13<br /> |Andy Zhang|| || |||| <br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/3/14<br /> |Andy Zhang|| 9:30||18:30 ||8|| <br /> *try to fix the bug<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/3/15<br /> |Andy Zhang|| 9:30||18:30 ||8|| <br /> *try to fix the bug<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/3/16<br /> |Andy Zhang|| 9:30||18:30 ||8|| <br /> *try to fix the bug<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/3/17<br /> |Andy Zhang|| 9:30||18:30 ||8|| <br /> *try to fix the bug<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/3/20<br /> |Andy Zhang|| 9:30||18:30 ||8|| <br /> *try to fix the bug<br /> *turns out the bug rises from the unfamiliarity with numpy.resize() function<br /> *the model can now deal with oov problem<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/3/21<br /> |Andy Zhang|| 9:30||18:30 ||8|| <br /> *working on graduation design document<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/3/22<br /> |Andy Zhang|| 9:30||18:30 ||8|| <br /> *working on graduation design document<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/3/23<br /> |Andy Zhang|| 9:30||18:30 ||8|| <br /> *working on graduation design document<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/3/24<br /> |Andy Zhang|| 9:30||18:30 ||8|| <br /> *working on graduation design document<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/3/27<br /> |Andy Zhang|| 9:30||18:30 ||8|| <br /> *rerun cs2en data on theano<br /> *run wmt17 zh2en data on theano<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/3/28<br /> |Andy Zhang|| 9:30||18:30 ||8|| <br /> *run zh2en data from Lihang's on theano<br /> *test the best bleu models on test set<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/3/29<br /> |Andy Zhang|| 9:30||18:30 ||8|| <br /> *rerun experiments as on Monday<br /> *tested results<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/3/30<br /> |Andy Zhang|||| |||| <br /> <br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2017/3/31<br /> |Andy Zhang|||| |||| <br /> <br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |}</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/Schedule 2017-04-02T01:09:52Z

<p>Zhangandy：/* Past progress */</p> <hr /> <div>=NLP Schedule=<br /> <br /> ==Members==<br /> <br /> ===Current Members===<br /> <br /> * Yang Feng (冯洋)<br /> * Jiyuan Zhang （张记袁）<br /> * Aodong Li (李傲冬)<br /> * Andi Zhang (张安迪)<br /> * Shiyue Zhang (张诗悦)<br /> * Li Gu (古丽)<br /> * Peilun Xiao (肖培伦)<br /> <br /> ===Former Members===<br /> * '''Chao Xing (邢超)''' : FreeNeb<br /> * '''Rong Liu (刘荣)''' : 优酷<br /> * '''Xiaoxi Wang (王晓曦)''' : 图灵机器人<br /> * '''Xi Ma (马习)''' : 清华大学研究生<br /> * '''Tianyi Luo (骆天一)''' ： phd candidate in University of California Santa Cruz<br /> * '''Qixin Wang (王琪鑫)''' : MA candidate in University of California<br /> * '''DongXu Zhang (张东旭)''': --<br /> * '''Yiqiao Pan (潘一桥)''' ： MA candidate in University of Sydney <br /> * '''Shiyao Li （李诗瑶）''' : BUPT<br /> * '''Aiting Liu (刘艾婷)''' : BUPT<br /> <br /> ==Work Progress==<br /> ===Daily Report===<br /> <br /> {|class=&quot;wikitable&quot;<br /> ! Date !! Person !! start!! leave !! hours ||status<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/5<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for ACL paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/6<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for ACL paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/7<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for ACL paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/8<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||11:30 || 20:00|| 6+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/9<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/10<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/11<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||14:30 || 20:00|| 5+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/13<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/14<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/15<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/16<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/17<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/18<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/19<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/21<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * found the reason of loss not go down<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/22<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:00 || 20:00|| 9+||<br /> * use cos to computer alignments, and loss can go down<br /> * replace original attention with cos attention, partly and fully train<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> | rowspan=&quot;3&quot;|2017/2/27<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang || 9:30 || 19:00|| 8+ ||<br /> * find the tanh linear problem<br /> * use 20*cos replace tanh, get 44.8 blue<br /> <br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;3&quot;|2017/2/28<br /> |Andy Zhang|| 12:00||19:00 || 7|| <br /> *read theano code of NMT, try to find out how it generated encoder input masks<br /> *did some coding of input masks on baseline_beam, need to be further tested<br /> |-<br /> |Shiyue Zhang || 9:30 || 19:00 || 8 +||<br /> * add encoding and attention mask <br /> * try to find the problem of mem attention<br /> <br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |}<br /> <br /> ===Time Off Table===<br /> <br /> {| class=&quot;wikitable&quot;<br /> ! Date !! Yang Feng !! Jiyuan Zhang <br /> |-<br /> |}<br /> <br /> ==Past progress==<br /> [[nlp-progress 2017/03]]<br /> <br /> [[nlp-progress 2017/02]]<br /> <br /> [[nlp-progress 2017/01]]<br /> <br /> [[nlp-progress 2016/12]]<br /> <br /> [[nlp-progress 2016/11]]<br /> <br /> [[nlp-progress 2016/10]]<br /> <br /> [[nlp-progress 2016/09]]<br /> <br /> [[nlp-progress 2016/08]]<br /> <br /> [[nlp-progress 2016/05/01 -- 08/16 | nlp-progress 2016/05-07]]<br /> <br /> [[nlp-progress 2016/04]]</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/Nlp-progress_2017/02 2017-04-02T01:09:05Z

<p>Zhangandy：以“===Daily Report=== {|class=&quot;wikitable&quot; ! Date !! Person !! start!! leave !! hours ||status |- | rowspan=&quot;4&quot;|2017/2/5 |Andy Zhang|| || || || |- |Shiyue Zhang ||9:3...”为内容创建页面</p> <hr /> <div>===Daily Report===<br /> <br /> {|class=&quot;wikitable&quot;<br /> ! Date !! Person !! start!! leave !! hours ||status<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/5<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for ACL paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/6<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for ACL paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/7<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for ACL paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/8<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||11:30 || 20:00|| 6+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/9<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/10<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/11<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||14:30 || 20:00|| 5+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/13<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/14<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/15<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/16<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/17<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/18<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/19<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/21<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * found the reason of loss not go down<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/22<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:00 || 20:00|| 9+||<br /> * use cos to computer alignments, and loss can go down<br /> * replace original attention with cos attention, partly and fully train<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> | rowspan=&quot;3&quot;|2017/2/27<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang || 9:30 || 19:00|| 8+ ||<br /> * find the tanh linear problem<br /> * use 20*cos replace tanh, get 44.8 blue<br /> <br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;3&quot;|2017/2/28<br /> |Andy Zhang|| 12:00||19:00 || 7|| <br /> *read theano code of NMT, try to find out how it generated encoder input masks<br /> *did some coding of input masks on baseline_beam, need to be further tested<br /> |-<br /> |Shiyue Zhang || 9:30 || 19:00 || 8 +||<br /> * add encoding and attention mask <br /> * try to find the problem of mem attention<br /> <br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |}</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/Schedule 2017-04-02T01:08:40Z

<p>Zhangandy：/* Past progress */</p> <hr /> <div>=NLP Schedule=<br /> <br /> ==Members==<br /> <br /> ===Current Members===<br /> <br /> * Yang Feng (冯洋)<br /> * Jiyuan Zhang （张记袁）<br /> * Aodong Li (李傲冬)<br /> * Andi Zhang (张安迪)<br /> * Shiyue Zhang (张诗悦)<br /> * Li Gu (古丽)<br /> * Peilun Xiao (肖培伦)<br /> <br /> ===Former Members===<br /> * '''Chao Xing (邢超)''' : FreeNeb<br /> * '''Rong Liu (刘荣)''' : 优酷<br /> * '''Xiaoxi Wang (王晓曦)''' : 图灵机器人<br /> * '''Xi Ma (马习)''' : 清华大学研究生<br /> * '''Tianyi Luo (骆天一)''' ： phd candidate in University of California Santa Cruz<br /> * '''Qixin Wang (王琪鑫)''' : MA candidate in University of California<br /> * '''DongXu Zhang (张东旭)''': --<br /> * '''Yiqiao Pan (潘一桥)''' ： MA candidate in University of Sydney <br /> * '''Shiyao Li （李诗瑶）''' : BUPT<br /> * '''Aiting Liu (刘艾婷)''' : BUPT<br /> <br /> ==Work Progress==<br /> ===Daily Report===<br /> <br /> {|class=&quot;wikitable&quot;<br /> ! Date !! Person !! start!! leave !! hours ||status<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/5<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for ACL paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/6<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for ACL paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/7<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for ACL paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/8<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||11:30 || 20:00|| 6+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/9<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/10<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/11<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||14:30 || 20:00|| 5+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/13<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/14<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/15<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/16<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/17<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/18<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/19<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/21<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * found the reason of loss not go down<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/22<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:00 || 20:00|| 9+||<br /> * use cos to computer alignments, and loss can go down<br /> * replace original attention with cos attention, partly and fully train<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> | rowspan=&quot;3&quot;|2017/2/27<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang || 9:30 || 19:00|| 8+ ||<br /> * find the tanh linear problem<br /> * use 20*cos replace tanh, get 44.8 blue<br /> <br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;3&quot;|2017/2/28<br /> |Andy Zhang|| 12:00||19:00 || 7|| <br /> *read theano code of NMT, try to find out how it generated encoder input masks<br /> *did some coding of input masks on baseline_beam, need to be further tested<br /> |-<br /> |Shiyue Zhang || 9:30 || 19:00 || 8 +||<br /> * add encoding and attention mask <br /> * try to find the problem of mem attention<br /> <br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |}<br /> <br /> ===Time Off Table===<br /> <br /> {| class=&quot;wikitable&quot;<br /> ! Date !! Yang Feng !! Jiyuan Zhang <br /> |-<br /> |}<br /> <br /> ==Past progress==<br /> [[nlp-progress 2017/02]]<br /> <br /> [[nlp-progress 2017/01]]<br /> <br /> [[nlp-progress 2016/12]]<br /> <br /> [[nlp-progress 2016/11]]<br /> <br /> [[nlp-progress 2016/10]]<br /> <br /> [[nlp-progress 2016/09]]<br /> <br /> [[nlp-progress 2016/08]]<br /> <br /> [[nlp-progress 2016/05/01 -- 08/16 | nlp-progress 2016/05-07]]<br /> <br /> [[nlp-progress 2016/04]]</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/NLP_Status_Report_2017-3-13 2017-03-14T01:46:23Z

<p>Zhangandy：</p> <hr /> <div>{| class=&quot;wikitable&quot;<br /> !Date !! People !! Last Week !! This Week<br /> |-<br /> | rowspan=&quot;6&quot;|2017/1/3<br /> |Yang Feng ||<br /> * tested and analyzed the results on the cs-en data set (30.4 on the heldout-training set and 7.3 on the dev set);<br /> * added masks to the baseline (44.4 on the cn-en);<br /> * added encoder-masks and memory-masks to alpha-gamma method and fixed the bugs. Got an improvement of 0.5 again the masked baseline [[http://cslt.riit.tsinghua.edu.cn/mediawiki/images/b/b8/Nmt_mn_report_continue.pdf report]];<br /> * To avoid doing softmax twice, rewrite the softmax_cross_entropy function myself. (under-training)<br /> ||<br /> * analyze and improve the alpha-gamma method.<br /> |-<br /> |Jiyuan Zhang ||<br /> *completed to reproduce planning neural network<br /> *chose best attention_memory model for huilian and ran big train dataset(about 370k) [http://cslt.riit.tsinghua.edu.cn/mediawiki/images/b/b9/Model_with_different_dataset.pdf result]<br /> <br /> || <br /> *Keyword expansion model<br /> *collect more poem from Internet<br /> *recruiting<br /> <br /> |-<br /> |Andi Zhang ||<br /> *ran baseline without mask, found that the model with masks has a slightly better bleu score.<br /> *tried a way to deal oov words; but it can't predict '_EOS' symbol<br /> ||<br /> *try to fix the problem<br /> |-<br /> |Shiyue Zhang || <br /> * added trained memory-attention model to neural model(43.0) and got 2+ blue gain (45.19), but need more validation and improvement<br /> * ran baseline model on cs-en data, and found it was good on train set but poor on test set.<br /> * ran baseline model on en-fr data, and found 'inf' problem.<br /> * fixed the 'inf' problem by debugging the code of mask-added baseline model.<br /> * running on cs-en and en-fr data again.<br /> ||<br /> * go on with baseline on big data: get results of cs-en and enfr data, train on zh-en data from [http://www.statmt.org/wmt17/translation-task.html#download WMT17]<br /> * go on to refine memory attention model: retrain to find out if the 2+ is just by chance, try more memory attention structure (relu, a(t-1), y(t-1)...)<br /> |-<br /> |Peilun Xiao ||<br /> <br /> ||<br /> <br /> |}</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/Schedule 2017-02-28T09:24:01Z

<p>Zhangandy：/* Daily Report */</p> <hr /> <div>=NLP Schedule=<br /> <br /> ==Members==<br /> <br /> ===Current Members===<br /> <br /> * Yang Feng (冯洋)<br /> * Jiyuan Zhang （张记袁）<br /> * Aodong Li (李傲冬)<br /> * Andi Zhang (张安迪)<br /> * Shiyue Zhang (张诗悦)<br /> * Li Gu (古丽)<br /> * Peilun Xiao (肖培伦)<br /> <br /> ===Former Members===<br /> * '''Chao Xing (邢超)''' : FreeNeb<br /> * '''Rong Liu (刘荣)''' : 优酷<br /> * '''Xiaoxi Wang (王晓曦)''' : 图灵机器人<br /> * '''Xi Ma (马习)''' : 清华大学研究生<br /> * '''Tianyi Luo (骆天一)''' ： phd candidate in University of California Santa Cruz<br /> * '''Qixin Wang (王琪鑫)''' : MA candidate in University of California<br /> * '''DongXu Zhang (张东旭)''': --<br /> * '''Yiqiao Pan (潘一桥)''' ： MA candidate in University of Sydney <br /> * '''Shiyao Li （李诗瑶）''' : BUPT<br /> * '''Aiting Liu (刘艾婷)''' : BUPT<br /> <br /> ==Work Progress==<br /> ===Daily Report===<br /> <br /> {|class=&quot;wikitable&quot;<br /> ! Date !! Person !! start!! leave !! hours ||status<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/5<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for ACL paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/6<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for ACL paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/7<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for ACL paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/8<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||11:30 || 20:00|| 6+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/9<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/10<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/11<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||14:30 || 20:00|| 5+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/13<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/14<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/15<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/16<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/17<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/18<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/19<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * prepare for IJCAI paper<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/21<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:30 || 20:00|| 9+||<br /> * found the reason of loss not go down<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> |-<br /> | rowspan=&quot;4&quot;|2017/2/22<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang ||9:00 || 20:00|| 9+||<br /> * use cos to computer alignments, and loss can go down<br /> * replace original attention with cos attention, partly and fully train<br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |Guli || || || ||<br /> |-<br /> | rowspan=&quot;3&quot;|2017/2/27<br /> |Andy Zhang|| || || || <br /> |-<br /> |Shiyue Zhang || || || ||<br /> <br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;3&quot;|2017/2/28<br /> |Andy Zhang|| 12:00||19:00 || 7|| <br /> *read theano code of NMT, try to find out how it generated encoder input masks<br /> *did some coding of input masks on baseline_beam, need to be further tested<br /> |-<br /> |Shiyue Zhang || || || ||<br /> <br /> |-<br /> |Peilun Xiao || || || ||<br /> |-<br /> |}<br /> <br /> ===Time Off Table===<br /> <br /> {| class=&quot;wikitable&quot;<br /> ! Date !! Yang Feng !! Jiyuan Zhang <br /> |-<br /> |}<br /> <br /> ==Past progress==<br /> [[nlp-progress 2017/01]]<br /> <br /> [[nlp-progress 2016/12]]<br /> <br /> [[nlp-progress 2016/11]]<br /> <br /> [[nlp-progress 2016/10]]<br /> <br /> [[nlp-progress 2016/09]]<br /> <br /> [[nlp-progress 2016/08]]<br /> <br /> [[nlp-progress 2016/05/01 -- 08/16 | nlp-progress 2016/05-07]]<br /> <br /> [[nlp-progress 2016/04]]</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/NLP_Status_Report_2017-1-10 2017-01-10T05:42:11Z

<p>Zhangandy：</p> <hr /> <div>{| class=&quot;wikitable&quot;<br /> !Date !! People !! Last Week !! This Week<br /> |-<br /> | rowspan=&quot;6&quot;|2017/1/3<br /> |Yang Feng ||<br /> *[[nmt+mn:]] tried to improve the nmt baseline;<br /> *met with problems for baseline, rulling out the factor of output order and file format and got the reason of learning rate.<br /> *read the code of Andy's;<br /> *wrote the code for bleu evaluation;<br /> *managed to fix the code of nmt+mn;<br /> *ran experiments [[http://cslt.riit.tsinghua.edu.cn/mediawiki/images/5/50/Nmt_mn_report.pdf report]]<br /> ||<br /> *[[nmt+mn:]] do further experiments.<br /> |-<br /> |Jiyuan Zhang ||<br /> *improved speed of prediction process<br /> *ran expriments:&lt;br/&gt;<br /> two sytles expriments of top1_memory_model&lt;br/&gt;<br /> overfitting expriments of top1_memory_model&lt;br/&gt;<br /> two styles expriments of average_memory_model&lt;br/&gt;<br /> overfitting expirments of average_memory_model<br /> || <br /> *improve poem model <br /> |-<br /> |Andi Zhang ||<br /> *helped Jiyuan dealing with poem questionnaires<br /> ||<br /> *continue this work, may start collecting feedback if all questionnaires are filled out<br /> |-<br /> |Shiyue Zhang || <br /> * finished running theano baseline <br /> * read and understood the beam search in theano baseline<br /> * started to write the [http://dynet-doc-zh.readthedocs.io/zh_CN/latest/ Dynet Chinese Document]<br /> * started to implement beam search to tensorflow baseline<br /> ||<br /> * finished the beam search implementation<br /> * go on with Dynet Chinese Document<br /> |-<br /> |Guli ||<br /> * run nmt with monolingual data<br /> * bleu computation<br /> * learn about tensorflow <br /> ||<br /> * improve my paper<br /> * analyze experiment results<br /> |-<br /> |Peilun Xiao ||<br /> *learned tf-idf algorithm<br /> *coded tf-idf alogrithm in python,but found it not worked well <br /> *tried to use small dataset to test the program<br /> ||<br /> *use sklearn tfidf to test the dataset<br /> |}</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/Bi-monthly-2016-12-language_Language_group 2017-01-04T04:03:21Z

<p>Zhangandy：以“Team [] Zhangandy [http://cslt.riit.tsinghua.edu.cn/mediawiki/images/c/c1/Bimonthly_andi_11-12.pdf]”为内容创建页面</p> <hr /> <div>Team []<br /> <br /> Zhangandy [http://cslt.riit.tsinghua.edu.cn/mediawiki/images/c/c1/Bimonthly_andi_11-12.pdf]</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/%E6%96%87%E4%BB%B6:Bimonthly_andi_11-12.pdf 2017-01-04T04:03:07Z

<p>Zhangandy：</p> <hr /> <div></div>

Zhangandy http://index.cslt.org/mediawiki/index.php/%E6%96%87%E4%BB%B6:Bimonthly_andi_11-12.pptx 2017-01-04T04:01:54Z

<p>Zhangandy：</p> <hr /> <div></div>

Zhangandy http://index.cslt.org/mediawiki/index.php/NLP_Status_Report_2017-1-3 2017-01-03T05:50:23Z

<p>Zhangandy：</p> <hr /> <div>{| class=&quot;wikitable&quot;<br /> !Date !! People !! Last Week !! This Week<br /> |-<br /> | rowspan=&quot;6&quot;|2016/12/26<br /> |Yang Feng ||<br /> *[[nmt+mn:]] tried to improve the nmt baseline;<br /> *read the code of Andy's;<br /> *wrote the code for bleu evaluation;<br /> *finished the code of nmt+mn;<br /> *ran experiments;<br /> ||<br /> *[[nmt+mn:]] do further experiments.<br /> |-<br /> |Jiyuan Zhang ||<br /> *integrated tone_model to attention_model for insteading manul rule,but the effect wasn't good<br /> *replacing　all_pz rule with half_pz<br /> *token a classical Chinese as input,generated poem [http://cslt.riit.tsinghua.edu.cn/mediawiki/images/3/33/Story_input.pdf]<br /> || <br /> *improve poem model <br /> |-<br /> |Andi Zhang ||<br /> *handed in previous codes to Mrs.Feng<br /> *help Jiyuan gather poems about tianyuan<br /> ||<br /> *help Jiyuan with his work<br /> *gather more poems<br /> |-<br /> |Shiyue Zhang || <br /> * tried to add true action info when training gate, which got better results than no true actions, but still not very good.<br /> * tried different scale vectors, and found setting &gt;=-5000 is good<br /> * tried to change cos to only inner product, and inner product is better than cos.<br /> * [[http://cslt.riit.tsinghua.edu.cn/mediawiki/images/9/9f/RNNG%2Bmm_experiment_report.pdf report]]<br /> * 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]]<br /> * trying the joint training, which got a problem of optimization. <br /> ||<br /> * try the joint training <br /> * read more papers and write a summary<br /> |-<br /> |Guli ||<br /> * finished the first draft of the survey<br /> * voice tagging <br /> ||<br /> * morpheme-based nmt<br /> * improve nmt with monolingual data<br /> |-<br /> |Peilun Xiao ||<br /> *learned tf-idf algorithm<br /> *coded tf-idf alogrithm in python,but found it not worked well <br /> *tried to use small dataset to test the program<br /> ||<br /> *use sklearn tfidf to test the dataset<br /> |}</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/Schedule 2017-01-02T08:03:18Z

<p>Zhangandy：/* Daily Report */</p> <hr /> <div>=NLP Schedule=<br /> <br /> ==Members==<br /> <br /> ===Current Members===<br /> <br /> * Yang Feng (冯洋)<br /> * Jiyuan Zhang （张记袁）<br /> * Aodong Li (李傲冬)<br /> * Andi Zhang (张安迪)<br /> * Shiyue Zhang (张诗悦)<br /> * Li Gu (古丽)<br /> * Peilun Xiao (肖培伦)<br /> <br /> ===Former Members===<br /> * '''Chao Xing (邢超)''' : FreeNeb<br /> * '''Rong Liu (刘荣)''' : 优酷<br /> * '''Xiaoxi Wang (王晓曦)''' : 图灵机器人<br /> * '''Xi Ma (马习)''' : 清华大学研究生<br /> * '''Tianyi Luo (骆天一)''' ： phd candidate in University of California Santa Cruz<br /> * '''Qixin Wang (王琪鑫)''' : MA candidate in University of California<br /> * '''DongXu Zhang (张东旭)''': --<br /> * '''Yiqiao Pan (潘一桥)''' ： MA candidate in University of Sydney <br /> * '''Shiyao Li （李诗瑶）''' : BUPT<br /> * '''Aiting Liu (刘艾婷)''' : BUPT<br /> <br /> ==Work Progress==<br /> ===Daily Report===<br /> <br /> {|class=&quot;wikitable&quot;<br /> ! Date !! Person !! start!! leave !! hours ||status<br /> |-<br /> | rowspan=&quot;3&quot;|2016/12/1<br /> |Andy Zhang|| 9:30||18:30 ||8 || <br /> *read source code<br /> *thought about ways to store encoder output<br /> |-<br /> |Shiyue Zhang || 9:10 || 19:00|| 8+ ||<br /> * draw tsne pictures<br /> |-<br /> |Guli ||9:10 ||9:50 ||12+ || <br /> *prepare data and run code<br /> *prepare for Thesis Report<br /> |- <br /> | rowspan=&quot;4&quot;|2016/12/2<br /> |Andy Zhang|| 9:30|| 18:30|| 8||<br /> *finish the code to output encoder output but have not tested yet<br /> *prepared a small data set for testing <br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||9:05 ||18:05 ||8 || <br /> *read code and check data <br /> *think about how to improve bleu score<br /> |- <br /> |Peilun Xiao ||10:00||12:30|| 2+||<br /> *learn basic things of work<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/5<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *group meeting<br /> *prepare the same data set as in Theano for Tensorflow<br /> |-<br /> |Shiyue Zhang || 9:30 || 19:00 || 8+||<br /> * discuss with teachers about the tsne pictures<br /> * group meeting<br /> |-<br /> |Guli ||9:00 ||22:35 ||13+ || <br /> *install moses<br /> *weekly report<br /> |- <br /> |Peilun Xiao ||9:30||18:30 ||8||<br /> *learn how to use weka<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/6<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *run fr-en (single encoder)<br /> *prepare paraphrase data set, enumerate them, excluding 呀啊哈, etc.<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||9:05 ||22;30 ||12+ || <br /> *run moses<br /> *write Thesis Report<br /> |- <br /> |Peilun Xiao ||9:40||18:30|| 8||<br /> *read a paper<br /> *use weka<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/7<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *prepare paraphrase data set, enumerate them, excluding 呀啊哈, etc.<br /> *met with some problems, used too much memory, had to restrict enumerate nums to 15000<br /> |-<br /> |Shiyue Zhang || 9:30 || 19:00 || 8+||<br /> * change mem to 28-dim states<br /> * retry previous experiments<br /> |-<br /> |Guli ||9:15||22;00||12+ || <br /> *run moses<br /> *write Thesis Report<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/8<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *finished preparing the data set<br /> *start working on bidirectional model <br /> |-<br /> |Shiyue Zhang || 9:30 || 19:00 || 8+||<br /> * try more experiments, but all are not good, always converge to baseline<br /> * draw the tsne picture of test data<br /> * discuss with Teacher Wang<br /> |-<br /> |Guli ||9:15||21;15||12 || <br /> *run moses<br /> *write Thesis Report<br /> |- <br /> |Peilun Xiao ||9:30||17:30|| 7||<br /> *learn GibbsLda++<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/9<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *finished coding, added masks for encoder inputs<br /> *had a some breakthrough with Jiyuan's help, but still meet nan problem <br /> |-<br /> |Shiyue Zhang ||9:30 || 19:00 || 8+||<br /> * return to 384-dim mem, which has 0.07 improvement than baseline<br /> * replace cos with a one-layer neural network, which is also slightly better than baseline<br /> * discuss with teachers, and find the problem of scale<br /> |-<br /> |Guli ||9:15||21;30||12 +|| <br /> *prepared Thesis Report<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/12<br /> |Andy Zhang|| || || ||<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||10:00||19;00|| 9|| <br /> *read paper to solve oov<br /> |- <br /> |Peilun Xiao ||9:30||19:00||8+||<br /> *use lda to generate document vector<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/13<br /> |Andy Zhang|| 9:30 || 18:30 || 8 ||<br /> *tried to modify the wrong softmax<br /> |-<br /> |Shiyue Zhang || 9:30 || 19:00 || 8+ ||<br /> * change mem scale to (0, -10000.0...)<br /> * try previous experiments<br /> |-<br /> |Guli ||09:30||21;00||11+|| <br /> *read papers<br /> |- <br /> |Peilun Xiao ||9:50 ||19:00 || 8 ||<br /> *use lda to generate document vector<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/14<br /> |Andy Zhang|| 9:30 || 18:30 || 8 ||<br /> *tried to modify the wrong softmax<br /> *decided to abandon it for now<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||09:20||20;30||11+|| <br /> *read papers<br /> *writing a survey<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/15<br /> |Andy Zhang|| 9:30 || 18:30 || 8 ||<br /> *tried to add bleu scoring into the code, but met with out of memory problem<br /> |-<br /> |Shiyue Zhang || 9:30 || 19:00|| 8+||<br /> * changed scale didn't lead to better results<br /> * try 1-dim gate, but converge to baseline<br /> |-<br /> |Guli ||09:20||20;30||11+|| <br /> *read papers<br /> *writing a survey<br /> |- <br /> |Peilun Xiao ||13:30||19:00||5+||<br /> *use lda to generate document vector<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/16<br /> |Andy Zhang|| 9:20 || 18:20 || 8 ||<br /> *finished adding bleu scoring into the code<br /> |-<br /> |Shiyue Zhang || 14:00 || 19:00|| 5||<br /> * try to only train gate <br /> * try model similar to attention<br /> |-<br /> |Guli ||09:30||20;30||11|| <br /> *Check the data, modify the error<br /> *writing a survey<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/19<br /> |Andy Zhang|| 9:30 || 18:30 || 8 ||<br /> *group meeting<br /> *made some revises on bleu scoring<br /> |-<br /> |Shiyue Zhang || 9:00 || 19:00|| 9||<br /> * review last week work<br /> * write the report and talk with teachers<br /> * decide what to do next and meeting<br /> * try to train gate with true action info<br /> |-<br /> |Guli ||9:15||20:00||10+|| <br /> *weekly meeting<br /> *conduct comparative test<br /> |- <br /> |Peilun Xiao ||9:30 ||19:00 || 8+ ||<br /> *weekly meeting<br /> *code on tf-idf <br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/20<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *dealt with some bugs on bidirectional model<br /> *coding on getting encoder outputs<br /> |-<br /> |Shiyue Zhang || 9:30 ||19:00|| 8+||<br /> * try to train gate with true actions, which gets the results better than no true action, but still not very good.<br /> * try to change cos to inner product, which gets better performance than cos.<br /> |-<br /> |Guli ||9:00||21:50||12+|| <br /> *voice tagging<br /> *write survey<br /> |- <br /> |Peilun Xiao ||9:50||18:30 ||8 ||<br /> *debug the code<br /> *prepare a small test set<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/21<br /> |Andy Zhang||9:20 || 18:30 || 8+ ||<br /> *finishing testing on getting encoder outputs<br /> *coded to output corresponding source and target sentences, but have not tested yet<br /> |-<br /> |Shiyue Zhang || || || ||<br /> <br /> |-<br /> |Guli ||9:10||21:10||12 || <br /> *voice tagging<br /> *write survey<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/22<br /> |Andy Zhang||9:20 || 18:30 || 8+ ||<br /> *decided to make some revises on bleu scoring part<br /> <br /> |-<br /> |Shiyue Zhang || 9:30 || 19:00|| 8+||<br /> * try to change the scales of one-hot vec, and find &gt;=-5000 is good<br /> |-<br /> |Guli ||9:20 ||21:10 ||11+ || <br /> *voice tagging<br /> *write survey<br /> |- <br /> |Peilun Xiao ||9:20 || 19:00|| 9 ||<br /> *read paper<br /> *try to solve the bug<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/23<br /> |Andy Zhang||9:30 || 18:10 || 7+ ||<br /> *finished bleu scoring part<br /> *made it process by batches &amp; testing several checkpoints and save the best one<br /> |-<br /> |Shiyue Zhang || 14:30 || 19:00|| 5.5 ||<br /> * review the work of the whole week and discuss with teachers<br /> * try joint training<br /> |-<br /> |Guli ||9:20|| 21:30 ||12+ || <br /> *voice tagging<br /> *write survey<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/26<br /> |Andy Zhang|| || || ||<br /> <br /> |-<br /> |Shiyue Zhang || 9:30 || 19:00|| 8+ ||<br /> * weekly report<br /> |-<br /> |Guli ||9:10 || 21:30 ||12+ || <br /> * weekly report<br /> * read pepar<br /> |- <br /> |Peilun Xiao || 9:20 ||19:00 ||8+ ||<br /> * debug code<br /> * read sklearn source code<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/27<br /> |Andy Zhang||9:30 || 18:30 || 8 ||<br /> *modified the BLEU script to compare translation result with original sentences<br /> |-<br /> |Shiyue Zhang || 9:20 || 14:00|| 5.5 ||<br /> * <br /> |- <br /> |Guli ||9:05||18:40 ||7+ || <br /> * read paper <br /> * prepare data<br /> |- <br /> |Peilun Xiao ||9:20||19:00 ||9 ||<br /> * debug code<br /> * write a code about elm<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/28<br /> |Andy Zhang||9:30 || 18:30 || 8 ||<br /> *modified BLEU script, now it is auto called while training<br /> |-<br /> |Shiyue Zhang || 9:20 || 19:00 || 8+ ||<br /> * modify mem to the states before dropout<br /> * update progress on CVSS<br /> |- <br /> |Guli ||9:30||15:30||6 || <br /> * bleu computation<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/29<br /> |Andy Zhang||9:30 || 18:30 || 8 ||<br /> *transferred to help Jiyuan with his work<br /> *gathered 70+ poems for trainning<br /> |-<br /> |Shiyue Zhang || 9:10|| 17:10 || 7.5 ||<br /> * pause the rnng work<br /> * start to help Teacher Feng with NMT<br /> |- <br /> |Guli ||9:30||9:00 ||11+ || <br /> * prepare and run data<br /> |- <br /> |Peilun Xiao ||9:30 ||18:00 ||7+||<br /> * prepare data<br /> * run data<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/30<br /> |Andy Zhang|| || || ||<br /> <br /> |-<br /> |Shiyue Zhang || 9:30|| 19:10 || 8+ ||<br /> * run the theano NMT code successfully<br /> |- <br /> |Guli |||||||| <br /> |- <br /> |Peilun Xiao || || ||||<br /> <br /> |-<br /> |}<br /> <br /> ===Time Off Table===<br /> <br /> {| class=&quot;wikitable&quot;<br /> ! Date !! Yang Feng !! Jiyuan Zhang <br /> |-<br /> |}<br /> <br /> ==Past progress==<br /> <br /> [[nlp-progress 2016/11]]<br /> <br /> [[nlp-progress 2016/10]]<br /> <br /> [[nlp-progress 2016/09]]<br /> <br /> [[nlp-progress 2016/08]]<br /> <br /> [[nlp-progress 2016/05/01 -- 08/16 | nlp-progress 2016/05-07]]<br /> <br /> [[nlp-progress 2016/04]]</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/Schedule 2016-12-29T06:16:46Z

<p>Zhangandy：/* Daily Report */</p> <hr /> <div>=NLP Schedule=<br /> <br /> ==Members==<br /> <br /> ===Current Members===<br /> <br /> * Yang Feng (冯洋)<br /> * Jiyuan Zhang （张记袁）<br /> * Aodong Li (李傲冬)<br /> * Andi Zhang (张安迪)<br /> * Shiyue Zhang (张诗悦)<br /> * Li Gu (古丽)<br /> * Peilun Xiao (肖培伦)<br /> <br /> ===Former Members===<br /> * '''Chao Xing (邢超)''' : FreeNeb<br /> * '''Rong Liu (刘荣)''' : 优酷<br /> * '''Xiaoxi Wang (王晓曦)''' : 图灵机器人<br /> * '''Xi Ma (马习)''' : 清华大学研究生<br /> * '''Tianyi Luo (骆天一)''' ： phd candidate in University of California Santa Cruz<br /> * '''Qixin Wang (王琪鑫)''' : MA candidate in University of California<br /> * '''DongXu Zhang (张东旭)''': --<br /> * '''Yiqiao Pan (潘一桥)''' ： MA candidate in University of Sydney <br /> * '''Shiyao Li （李诗瑶）''' : BUPT<br /> * '''Aiting Liu (刘艾婷)''' : BUPT<br /> <br /> ==Work Progress==<br /> ===Daily Report===<br /> <br /> {|class=&quot;wikitable&quot;<br /> ! Date !! Person !! start!! leave !! hours ||status<br /> |-<br /> | rowspan=&quot;3&quot;|2016/12/1<br /> |Andy Zhang|| 9:30||18:30 ||8 || <br /> *read source code<br /> *thought about ways to store encoder output<br /> |-<br /> |Shiyue Zhang || 9:10 || 19:00|| 8+ ||<br /> * draw tsne pictures<br /> |-<br /> |Guli ||9:10 ||9:50 ||12+ || <br /> *prepare data and run code<br /> *prepare for Thesis Report<br /> |- <br /> | rowspan=&quot;4&quot;|2016/12/2<br /> |Andy Zhang|| 9:30|| 18:30|| 8||<br /> *finish the code to output encoder output but have not tested yet<br /> *prepared a small data set for testing <br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||9:05 ||18:05 ||8 || <br /> *read code and check data <br /> *think about how to improve bleu score<br /> |- <br /> |Peilun Xiao ||10:00||12:30|| 2+||<br /> *learn basic things of work<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/5<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *group meeting<br /> *prepare the same data set as in Theano for Tensorflow<br /> |-<br /> |Shiyue Zhang || 9:30 || 19:00 || 8+||<br /> * discuss with teachers about the tsne pictures<br /> * group meeting<br /> |-<br /> |Guli ||9:00 ||22:35 ||13+ || <br /> *install moses<br /> *weekly report<br /> |- <br /> |Peilun Xiao ||9:30||18:30 ||8||<br /> *learn how to use weka<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/6<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *run fr-en (single encoder)<br /> *prepare paraphrase data set, enumerate them, excluding 呀啊哈, etc.<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||9:05 ||22;30 ||12+ || <br /> *run moses<br /> *write Thesis Report<br /> |- <br /> |Peilun Xiao ||9:40||18:30|| 8||<br /> *read a paper<br /> *use weka<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/7<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *prepare paraphrase data set, enumerate them, excluding 呀啊哈, etc.<br /> *met with some problems, used too much memory, had to restrict enumerate nums to 15000<br /> |-<br /> |Shiyue Zhang || 9:30 || 19:00 || 8+||<br /> * change mem to 28-dim states<br /> * retry previous experiments<br /> |-<br /> |Guli ||9:15||22;00||12+ || <br /> *run moses<br /> *write Thesis Report<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/8<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *finished preparing the data set<br /> *start working on bidirectional model <br /> |-<br /> |Shiyue Zhang || 9:30 || 19:00 || 8+||<br /> * try more experiments, but all are not good, always converge to baseline<br /> * draw the tsne picture of test data<br /> * discuss with Teacher Wang<br /> |-<br /> |Guli ||9:15||21;15||12 || <br /> *run moses<br /> *write Thesis Report<br /> |- <br /> |Peilun Xiao ||9:30||17:30|| 7||<br /> *learn GibbsLda++<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/9<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *finished coding, added masks for encoder inputs<br /> *had a some breakthrough with Jiyuan's help, but still meet nan problem <br /> |-<br /> |Shiyue Zhang ||9:30 || 19:00 || 8+||<br /> * return to 384-dim mem, which has 0.07 improvement than baseline<br /> * replace cos with a one-layer neural network, which is also slightly better than baseline<br /> * discuss with teachers, and find the problem of scale<br /> |-<br /> |Guli ||9:15||21;30||12 +|| <br /> *prepared Thesis Report<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/12<br /> |Andy Zhang|| || || ||<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||10:00||19;00|| 9|| <br /> *read paper to solve oov<br /> |- <br /> |Peilun Xiao ||9:30||19:00||9+||<br /> *use lda to generate document vector<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/13<br /> |Andy Zhang|| 9:30 || 18:30 || 8 ||<br /> *tried to modify the wrong softmax<br /> |-<br /> |Shiyue Zhang || 9:30 || 19:00 || 8+ ||<br /> * change mem scale to (0, -10000.0...)<br /> * try previous experiments<br /> |-<br /> |Guli ||09:30||21;00||11+|| <br /> *read papers<br /> |- <br /> |Peilun Xiao ||9:50 ||19:00 || 9 ||<br /> *use lda to generate document vector<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/14<br /> |Andy Zhang|| 9:30 || 18:30 || 8 ||<br /> *tried to modify the wrong softmax<br /> *decided to abandon it for now<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||09:20||20;30||11+|| <br /> *read papers<br /> *writing a survey<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/15<br /> |Andy Zhang|| 9:30 || 18:30 || 8 ||<br /> *tried to add bleu scoring into the code, but met with out of memory problem<br /> |-<br /> |Shiyue Zhang || 9:30 || 19:00|| 8+||<br /> * changed scale didn't lead to better results<br /> * try 1-dim gate, but converge to baseline<br /> |-<br /> |Guli ||09:20||20;30||11+|| <br /> *read papers<br /> *writing a survey<br /> |- <br /> |Peilun Xiao ||13:30||19:00||5+||<br /> *use lda to generate document vector<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/16<br /> |Andy Zhang|| 9:20 || 18:20 || 8 ||<br /> *finished adding bleu scoring into the code<br /> |-<br /> |Shiyue Zhang || 14:00 || 19:00|| 5||<br /> * try to only train gate <br /> * try model similar to attention<br /> |-<br /> |Guli ||09:30||20;30||11|| <br /> *Check the data, modify the error<br /> *writing a survey<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/19<br /> |Andy Zhang|| 9:30 || 18:30 || 8 ||<br /> *group meeting<br /> *made some revises on bleu scoring<br /> |-<br /> |Shiyue Zhang || 9:00 || 19:00|| 9||<br /> * review last week work<br /> * write the report and talk with teachers<br /> * decide what to do next and meeting<br /> * try to train gate with true action info<br /> |-<br /> |Guli ||9：15||20：00||10+|| <br /> *weekly meeting<br /> *conduct comparative test<br /> |- <br /> |Peilun Xiao ||9：30 ||19：00 || 9+ ||<br /> *weekly meeting<br /> *code on tf-idf <br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/20<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *dealt with some bugs on bidirectional model<br /> *coding on getting encoder outputs<br /> |-<br /> |Shiyue Zhang || 9:30 ||19:00|| 8+||<br /> * try to train gate with true actions, which gets the results better than no true action, but still not very good.<br /> * try to change cos to inner product, which gets better performance than cos.<br /> |-<br /> |Guli ||9：00||21：50||12+|| <br /> *voice tagging<br /> *write survey<br /> |- <br /> |Peilun Xiao ||9:50||18:30 ||8 ||<br /> *debug the code<br /> *prepare a small test set<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/21<br /> |Andy Zhang||9:20 || 18:30 || 8+ ||<br /> *finishing testing on getting encoder outputs<br /> *coded to output corresponding source and target sentences, but have not tested yet<br /> |-<br /> |Shiyue Zhang || || || ||<br /> <br /> |-<br /> |Guli ||9：10||21：10 ||12 || <br /> *voice tagging<br /> *write survey<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/22<br /> |Andy Zhang||9:20 || 18:30 || 8+ ||<br /> *decided to make some revises on bleu scoring part<br /> <br /> |-<br /> |Shiyue Zhang || 9:30 || 19:00|| 8+||<br /> * try to change the scales of one-hot vec, and find &gt;=-5000 is good<br /> |-<br /> |Guli ||9：20 ||21:10 ||11+ || <br /> *voice tagging<br /> *write survey<br /> |- <br /> |Peilun Xiao ||9:20 || 19:00|| 9 ||<br /> *read paper<br /> *try to solve the bug<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/23<br /> |Andy Zhang||9:30 || 18:10 || 7+ ||<br /> *finished bleu scoring part<br /> *made it process by batches &amp; testing several checkpoints and save the best one<br /> |-<br /> |Shiyue Zhang || 14:30 || 19:00|| 5.5 ||<br /> * review the work of the whole week and discuss with teachers<br /> * try joint training<br /> |-<br /> |Guli ||9：20|| 21:30 ||12+ || <br /> *voice tagging<br /> *write survey<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/26<br /> |Andy Zhang|| || || ||<br /> <br /> |-<br /> |Shiyue Zhang || 14:30 || 19:00|| 5.5 ||<br /> * <br /> |-<br /> |Guli ||9:10 || 21:30 ||12+ || <br /> * weekly report<br /> * read pepar<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/27<br /> |Andy Zhang||9:30 || 18:30 || 8 ||<br /> *modified the BLEU script to compare translation result with original sentences<br /> |-<br /> |Shiyue Zhang || 14:30 || 19:00|| 5.5 ||<br /> *<br /> |- <br /> |Guli ||9：05||18:40 ||7+ || <br /> * read paper <br /> * prepare data<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/28<br /> |Andy Zhang||9:30 || 18:30 || 8 ||<br /> *modified BLEU script, now it is auto called while training<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |- <br /> |Guli |||| || || <br /> <br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/29<br /> |Andy Zhang||9:30 || 18:30 || 8 ||<br /> <br /> |-<br /> |Shiyue Zhang || || || ||<br /> <br /> |- <br /> |Guli |||| || || <br /> <br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> |}<br /> <br /> ===Time Off Table===<br /> <br /> {| class=&quot;wikitable&quot;<br /> ! Date !! Yang Feng !! Jiyuan Zhang <br /> |-<br /> |}<br /> <br /> ==Past progress==<br /> <br /> [[nlp-progress 2016/11]]<br /> <br /> [[nlp-progress 2016/10]]<br /> <br /> [[nlp-progress 2016/09]]<br /> <br /> [[nlp-progress 2016/08]]<br /> <br /> [[nlp-progress 2016/05/01 -- 08/16 | nlp-progress 2016/05-07]]<br /> <br /> [[nlp-progress 2016/04]]</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/NLP_Status_Report_2016-12-26 2016-12-26T05:00:55Z

<p>Zhangandy：</p> <hr /> <div>{| class=&quot;wikitable&quot;<br /> !Date !! People !! Last Week !! This Week<br /> |-<br /> | rowspan=&quot;6&quot;|2016/12/26<br /> |Yang Feng ||<br /> *[[s2smn:]] read six papers to fix the details of our model;<br /> *wrote the proposal of lexical memory and discussed the details with Teach Wang;<br /> *finished coding of only adding attention to the decoder and under debugging;<br /> *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]] ;<br /> *prepare the dictionary for the memory loading;<br /> *[[Huilan:]] documentation<br /> ||<br /> *[[s2smn:]] run the ecperiments.<br /> *[[rnng+mn:]] try to find the problem.<br /> |-<br /> |Jiyuan Zhang ||<br /> *coded tone_model,but had some trouble<br /> *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<br /> || <br /> *improve poem model <br /> |-<br /> |Andi Zhang ||<br /> *coded to output encoder outputs and correspoding source &amp; target sentences(ids in dictionaries)<br /> *coded a script for bleu scoring, which tests the five checkpoints auto created by training process and save the one with best performance<br /> ||<br /> *<br /> |-<br /> |Shiyue Zhang || <br /> * tried to add true action info when training gate, which got better results than no true actions, but still not very good.<br /> * tried different scale vectors, and found setting &gt;=-5000 is good<br /> * tried to change cos to only inner product, and inner product is better than cos.<br /> * [[http://cslt.riit.tsinghua.edu.cn/mediawiki/images/9/9f/RNNG%2Bmm_experiment_report.pdf report]]<br /> * 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]]<br /> * trying the joint training, which got a problem of optimization. <br /> ||<br /> * try the joint training <br /> * read more papers and write a summary<br /> |-<br /> |Guli ||<br /> *read papers about Transfer learning and solving OOV<br /> *conducted comparative test <br /> *writing survey<br /> ||<br /> * complete the first draft of the survey <br /> |-<br /> |Peilun Xiao ||<br /> *learned tf-idf algorithm<br /> *coded tf-idf alogrithm in python,but found it not worked well <br /> *tried to use small dataset to test the program<br /> ||<br /> *use sklearn tfidf to test the dataset<br /> |}</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/NLP_Status_Report_2016-12-19 2016-12-26T00:58:07Z

<p>Zhangandy：</p> <hr /> <div>{| class=&quot;wikitable&quot;<br /> !Date !! People !! Last Week !! This Week<br /> |-<br /> | rowspan=&quot;6&quot;|2016/12/19<br /> |Yang Feng ||<br /> *[[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]]<br /> *wrote part of the code of mn.<br /> *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]]<br /> *[[Huilan:]] fixed the problem of syntax-based translation.<br /> *sort out the system and corresponding documents.<br /> ||<br /> *[[s2smn:]] finish the code of adding mn.<br /> *[[Huilan:]] handover.<br /> |-<br /> |Jiyuan Zhang ||<br /> *coded tone_model,but had some trouble<br /> *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<br /> || <br /> *improve poem model <br /> |-<br /> |Andi Zhang ||<br /> *coded to output encoder outputs and correspoding source &amp; target sentences(ids in dictionaries)<br /> *coded a script for bleu scoring, which tests the five checkpoints auto created by training process and save the one with best performance<br /> ||<br /> *extract encoder outputs<br /> |-<br /> |Shiyue Zhang || <br /> * changed the one-hot vector to (0, -inf, -inf...), and retied the experiments. But no improvement showed.<br /> * tried 1-dim gate, but converged to baseline<br /> * tried to only train gate, but the best is taking all instance as &quot;right&quot;<br /> * trying a model similar to attention<br /> * [[http://cslt.riit.tsinghua.edu.cn/mediawiki/images/9/9f/RNNG%2Bmm_experiment_report.pdf report]]<br /> ||<br /> * try to add true action info when training gate<br /> * try different scale vectors<br /> * try to change cos to only inner product<br /> |-<br /> |Guli ||<br /> *read papers about Transfer learning and solving OOV<br /> *conducted comparative test <br /> *writing survey<br /> ||<br /> * complete the first draft of the survey <br /> |-<br /> |Peilun Xiao ||<br /> *use LDA to generate 10-500 dimension document vector in the rest datasets<br /> *write a python code about a new algorithm about tf-idf<br /> ||<br /> *debug the code<br /> |}</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/Schedule 2016-12-26T00:53:45Z

<p>Zhangandy：/* Daily Report */</p> <hr /> <div>=NLP Schedule=<br /> <br /> ==Members==<br /> <br /> ===Current Members===<br /> <br /> * Yang Feng (冯洋)<br /> * Jiyuan Zhang （张记袁）<br /> * Aodong Li (李傲冬)<br /> * Andi Zhang (张安迪)<br /> * Shiyue Zhang (张诗悦)<br /> * Li Gu (古丽)<br /> * Peilun Xiao (肖培伦)<br /> <br /> ===Former Members===<br /> * '''Chao Xing (邢超)''' : FreeNeb<br /> * '''Rong Liu (刘荣)''' : 优酷<br /> * '''Xiaoxi Wang (王晓曦)''' : 图灵机器人<br /> * '''Xi Ma (马习)''' : 清华大学研究生<br /> * '''Tianyi Luo (骆天一)''' ： phd candidate in University of California Santa Cruz<br /> * '''Qixin Wang (王琪鑫)''' : MA candidate in University of California<br /> * '''DongXu Zhang (张东旭)''': --<br /> * '''Yiqiao Pan (潘一桥)''' ： MA candidate in University of Sydney <br /> * '''Shiyao Li （李诗瑶）''' : BUPT<br /> * '''Aiting Liu (刘艾婷)''' : BUPT<br /> <br /> ==Work Progress==<br /> ===Daily Report===<br /> <br /> {|class=&quot;wikitable&quot;<br /> ! Date !! Person !! start!! leave !! hours ||status<br /> |-<br /> | rowspan=&quot;3&quot;|2016/12/1<br /> |Andy Zhang|| 9:30||18:30 ||8 || <br /> *read source code<br /> *thought about ways to store encoder output<br /> |-<br /> |Shiyue Zhang || 9:10 || 19:00|| 8+ ||<br /> * draw tsne pictures<br /> |-<br /> |Guli ||9:10 ||9:50 ||12+ || <br /> *prepare data and run code<br /> *prepare for Thesis Report<br /> |- <br /> | rowspan=&quot;4&quot;|2016/12/2<br /> |Andy Zhang|| 9:30|| 18:30|| 8||<br /> *finish the code to output encoder output but have not tested yet<br /> *prepared a small data set for testing <br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||9:05 ||18:05 ||8 || <br /> *read code and check data <br /> *think about how to improve bleu score<br /> |- <br /> |Peilun Xiao ||10:00||12:30|| 2+||<br /> *learn basic things of work<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/5<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *group meeting<br /> *prepare the same data set as in Theano for Tensorflow<br /> |-<br /> |Shiyue Zhang || 9:30 || 19:00 || 8+||<br /> * discuss with teachers about the tsne pictures<br /> * group meeting<br /> |-<br /> |Guli ||9:00 ||22:35 ||13+ || <br /> *install moses<br /> *weekly report<br /> |- <br /> |Peilun Xiao ||9:30||18:30 ||8||<br /> *learn how to use weka<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/6<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *run fr-en (single encoder)<br /> *prepare paraphrase data set, enumerate them, excluding 呀啊哈, etc.<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||9:05 ||22;30 ||12+ || <br /> *run moses<br /> *write Thesis Report<br /> |- <br /> |Peilun Xiao ||9:40||18:30|| 8||<br /> *read a paper<br /> *use weka<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/7<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *prepare paraphrase data set, enumerate them, excluding 呀啊哈, etc.<br /> *met with some problems, used too much memory, had to restrict enumerate nums to 15000<br /> |-<br /> |Shiyue Zhang || 9:30 || 19:00 || 8+||<br /> * change mem to 28-dim states<br /> * retry previous experiments<br /> |-<br /> |Guli ||9:15||22;00||12+ || <br /> *run moses<br /> *write Thesis Report<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/8<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *finished preparing the data set<br /> *start working on bidirectional model <br /> |-<br /> |Shiyue Zhang || 9:30 || 19:00 || 8+||<br /> * try more experiments, but all are not good, always converge to baseline<br /> * draw the tsne picture of test data<br /> * discuss with Teacher Wang<br /> |-<br /> |Guli ||9:15||21;15||12 || <br /> *run moses<br /> *write Thesis Report<br /> |- <br /> |Peilun Xiao ||9:30||17:30|| 7||<br /> *learn GibbsLda++<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/9<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *finished coding, added masks for encoder inputs<br /> *had a some breakthrough with Jiyuan's help, but still meet nan problem <br /> |-<br /> |Shiyue Zhang ||9:30 || 19:00 || 8+||<br /> * return to 384-dim mem, which has 0.07 improvement than baseline<br /> * replace cos with a one-layer neural network, which is also slightly better than baseline<br /> * discuss with teachers, and find the problem of scale<br /> |-<br /> |Guli ||9:15||21;30||12 +|| <br /> *prepared Thesis Report<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/12<br /> |Andy Zhang|| || || ||<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||10:00||19;00|| 9|| <br /> *read paper to solve oov<br /> |- <br /> |Peilun Xiao ||9:30||19:00||9+||<br /> *use lda to generate document vector<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/13<br /> |Andy Zhang|| 9:30 || 18:30 || 8 ||<br /> *tried to modify the wrong softmax<br /> |-<br /> |Shiyue Zhang || 9:30 || 19:00 || 8+ ||<br /> * change mem scale to (0, -10000.0...)<br /> * try previous experiments<br /> |-<br /> |Guli ||09:30||21;00||11+|| <br /> *read papers<br /> |- <br /> |Peilun Xiao ||9:50 ||19:00 || 9 ||<br /> *use lda to generate document vector<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/14<br /> |Andy Zhang|| 9:30 || 18:30 || 8 ||<br /> *tried to modify the wrong softmax<br /> *decided to abandon it for now<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||09:20||20;30||11+|| <br /> *read papers<br /> *writing a survey<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/15<br /> |Andy Zhang|| 9:30 || 18:30 || 8 ||<br /> *tried to add bleu scoring into the code, but met with out of memory problem<br /> |-<br /> |Shiyue Zhang || 9:30 || 19:00|| 8+||<br /> * changed scale didn't lead to better results<br /> * try 1-dim gate, but converge to baseline<br /> |-<br /> |Guli ||09:20||20;30||11+|| <br /> *read papers<br /> *writing a survey<br /> |- <br /> |Peilun Xiao ||13:30||19:00||5+||<br /> *use lda to generate document vector<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/16<br /> |Andy Zhang|| 9:20 || 18:20 || 8 ||<br /> *finished adding bleu scoring into the code<br /> |-<br /> |Shiyue Zhang || 14:00 || 19:00|| 5||<br /> * try to only train gate <br /> * try model similar to attention<br /> |-<br /> |Guli ||09:30||20;30||11|| <br /> *Check the data, modify the error<br /> *writing a survey<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/19<br /> |Andy Zhang|| 9:30 || 18:30 || 8 ||<br /> *group meeting<br /> *made some revises on bleu scoring<br /> |-<br /> |Shiyue Zhang || 9:00 || 19:00|| 9||<br /> * review last week work<br /> * write the report and talk with teachers<br /> * decide what to do next and meeting<br /> * try to train gate with true action info<br /> |-<br /> |Guli ||9：15||20：00||10+|| <br /> *weekly meeting<br /> *conduct comparative test<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/20<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *dealt with some bugs on bidirectional model<br /> *coding on getting encoder outputs<br /> |-<br /> |Shiyue Zhang || |||| ||<br /> <br /> |-<br /> |Guli ||9：00||21：50||12+|| <br /> *voice tagging<br /> *write survey<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/21<br /> |Andy Zhang||9:20 || 18:30 || 8+ ||<br /> *finishing testing on getting encoder outputs<br /> *coded to output corresponding source and target sentences, but have not tested yet<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||9：10||21：10 ||12 || <br /> *voice tagging<br /> *write survey<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/22<br /> |Andy Zhang||9:20 || 18:30 || 8+ ||<br /> *decided to make some revises on bleu scoring part<br /> <br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||9：20 ||21:10 ||11+ || <br /> *voice tagging<br /> *write survey<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/23<br /> |Andy Zhang||9:30 || 18:10 || 7+ ||<br /> *finished bleu scoring part<br /> *made it process by batches &amp; testing several checkpoints and save the best one<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||9：20|| 21:30 ||12+ || <br /> *voice tagging<br /> *write survey<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> |}<br /> <br /> ===Time Off Table===<br /> <br /> {| class=&quot;wikitable&quot;<br /> ! Date !! Yang Feng !! Jiyuan Zhang <br /> |-<br /> |}<br /> <br /> ==Past progress==<br /> <br /> [[nlp-progress 2016/11]]<br /> <br /> [[nlp-progress 2016/10]]<br /> <br /> [[nlp-progress 2016/09]]<br /> <br /> [[nlp-progress 2016/08]]<br /> <br /> [[nlp-progress 2016/05/01 -- 08/16 | nlp-progress 2016/05-07]]<br /> <br /> [[nlp-progress 2016/04]]</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/Schedule 2016-12-21T09:13:31Z

<p>Zhangandy：/* Daily Report */</p> <hr /> <div>=NLP Schedule=<br /> <br /> ==Members==<br /> <br /> ===Current Members===<br /> <br /> * Yang Feng (冯洋)<br /> * Jiyuan Zhang （张记袁）<br /> * Aodong Li (李傲冬)<br /> * Andi Zhang (张安迪)<br /> * Shiyue Zhang (张诗悦)<br /> * Li Gu (古丽)<br /> * Peilun Xiao (肖培伦)<br /> <br /> ===Former Members===<br /> * '''Chao Xing (邢超)''' : FreeNeb<br /> * '''Rong Liu (刘荣)''' : 优酷<br /> * '''Xiaoxi Wang (王晓曦)''' : 图灵机器人<br /> * '''Xi Ma (马习)''' : 清华大学研究生<br /> * '''Tianyi Luo (骆天一)''' ： phd candidate in University of California Santa Cruz<br /> * '''Qixin Wang (王琪鑫)''' : MA candidate in University of California<br /> * '''DongXu Zhang (张东旭)''': --<br /> * '''Yiqiao Pan (潘一桥)''' ： MA candidate in University of Sydney <br /> * '''Shiyao Li （李诗瑶）''' : BUPT<br /> * '''Aiting Liu (刘艾婷)''' : BUPT<br /> <br /> ==Work Progress==<br /> ===Daily Report===<br /> <br /> {|class=&quot;wikitable&quot;<br /> ! Date !! Person !! start!! leave !! hours ||status<br /> |-<br /> | rowspan=&quot;3&quot;|2016/12/1<br /> |Andy Zhang|| 9:30||18:30 ||8 || <br /> *read source code<br /> *thought about ways to store encoder output<br /> |-<br /> |Shiyue Zhang || 9:10 || 19:00|| 8+ ||<br /> * draw tsne pictures<br /> |-<br /> |Guli ||9:10 ||9:50 ||12+ || <br /> *prepare data and run code<br /> *prepare for Thesis Report<br /> |- <br /> | rowspan=&quot;4&quot;|2016/12/2<br /> |Andy Zhang|| 9:30|| 18:30|| 8||<br /> *finish the code to output encoder output but have not tested yet<br /> *prepared a small data set for testing <br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||9:05 ||18:05 ||8 || <br /> *read code and check data <br /> *think about how to improve bleu score<br /> |- <br /> |Peilun Xiao ||10:00||12:30|| 2+||<br /> *learn basic things of work<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/5<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *group meeting<br /> *prepare the same data set as in Theano for Tensorflow<br /> |-<br /> |Shiyue Zhang || 9:30 || 19:00 || 8+||<br /> * discuss with teachers about the tsne pictures<br /> * group meeting<br /> |-<br /> |Guli ||9:00 ||22:35 ||13+ || <br /> *install moses<br /> *weekly report<br /> |- <br /> |Peilun Xiao ||9:30||18:30 ||8||<br /> *learn how to use weka<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/6<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *run fr-en (single encoder)<br /> *prepare paraphrase data set, enumerate them, excluding 呀啊哈, etc.<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||9:05 ||22;30 ||12+ || <br /> *run moses<br /> *write Thesis Report<br /> |- <br /> |Peilun Xiao ||9:40||18:30|| 8||<br /> *read a paper<br /> *use weka<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/7<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *prepare paraphrase data set, enumerate them, excluding 呀啊哈, etc.<br /> *met with some problems, used too much memory, had to restrict enumerate nums to 15000<br /> |-<br /> |Shiyue Zhang || 9:30 || 19:00 || 8+||<br /> * change mem to 28-dim states<br /> * retry previous experiments<br /> |-<br /> |Guli ||9:15||22;00||12+ || <br /> *run moses<br /> *write Thesis Report<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/8<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *finished preparing the data set<br /> *start working on bidirectional model <br /> |-<br /> |Shiyue Zhang || 9:30 || 19:00 || 8+||<br /> * try more experiments, but all are not good, always converge to baseline<br /> * draw the tsne picture of test data<br /> * discuss with Teacher Wang<br /> |-<br /> |Guli ||9:15||21;15||12 || <br /> *run moses<br /> *write Thesis Report<br /> |- <br /> |Peilun Xiao ||9:30||17:30|| 7||<br /> *learn GibbsLda++<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/9<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *finished coding, added masks for encoder inputs<br /> *had a some breakthrough with Jiyuan's help, but still meet nan problem <br /> |-<br /> |Shiyue Zhang ||9:30 || 19:00 || 8+||<br /> * return to 384-dim mem, which has 0.07 improvement than baseline<br /> * replace cos with a one-layer neural network, which is also slightly better than baseline<br /> * discuss with teachers, and find the problem of scale<br /> |-<br /> |Guli ||9:15||21;30||12 +|| <br /> *prepared Thesis Report<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/12<br /> |Andy Zhang|| || || ||<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||10:00||19;00|| 9|| <br /> *read paper to solve oov<br /> |- <br /> |Peilun Xiao ||9:30||19:00||9+||<br /> *use lda to generate document vector<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/13<br /> |Andy Zhang|| 9:30 || 18:30 || 8 ||<br /> *tried to modify the wrong softmax<br /> |-<br /> |Shiyue Zhang || 9:30 || 19:00 || 8+ ||<br /> * change mem scale to (0, -10000.0...)<br /> * try previous experiments<br /> |-<br /> |Guli ||09:30||21;00||11+|| <br /> *read papers<br /> |- <br /> |Peilun Xiao ||9:50 ||19:00 || 9 ||<br /> *use lda to generate document vector<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/14<br /> |Andy Zhang|| 9:30 || 18:30 || 8 ||<br /> *tried to modify the wrong softmax<br /> *decided to abandon it for now<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||09:20||20;30||11+|| <br /> *read papers<br /> *writing a survey<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/15<br /> |Andy Zhang|| 9:30 || 18:30 || 8 ||<br /> *tried to add bleu scoring into the code, but met with out of memory problem<br /> |-<br /> |Shiyue Zhang || 9:30 || 19:00|| 8+||<br /> * changed scale didn't lead to better results<br /> * try 1-dim gate, but converge to baseline<br /> |-<br /> |Guli ||09:20||20;30||11+|| <br /> *read papers<br /> *writing a survey<br /> |- <br /> |Peilun Xiao ||13:30||19:00||5+||<br /> *use lda to generate document vector<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/16<br /> |Andy Zhang|| 9:20 || 18:20 || 8 ||<br /> *finished adding bleu scoring into the code<br /> |-<br /> |Shiyue Zhang || 14:00 || 19:00|| 5||<br /> * try to only train gate <br /> * try model similar to attention<br /> |-<br /> |Guli ||09:30||20;30||11|| <br /> *Check the data, modify the error<br /> *writing a survey<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/19<br /> |Andy Zhang|| 9:30 || 18:30 || 8 ||<br /> *group meeting<br /> *made some revises on bleu scoring<br /> |-<br /> |Shiyue Zhang || 9:00 || 19:00|| 9||<br /> * review last week work<br /> * write the report and talk with teachers<br /> * decide what to do next and meeting<br /> * try to train gate with true action info<br /> |-<br /> |Guli ||9：15||20：00||10+|| <br /> *weekly meeting<br /> *conduct comparative test<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/20<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *dealt with some bugs on bidirectional model<br /> *coding on getting encoder outputs<br /> |-<br /> |Shiyue Zhang || |||| ||<br /> <br /> |-<br /> |Guli ||9：00||21：50||12+|| <br /> *voice tagging<br /> *write survey<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/21<br /> |Andy Zhang||9:20 || 18:30 || 8+ ||<br /> *finishing testing on getting encoder outputs<br /> *coded to output corresponding source and target sentences, but have not tested yet<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||9：10|| || || <br /> *voice tagging<br /> *write survey<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> |}<br /> <br /> ===Time Off Table===<br /> <br /> {| class=&quot;wikitable&quot;<br /> ! Date !! Yang Feng !! Jiyuan Zhang <br /> |-<br /> |}<br /> <br /> ==Past progress==<br /> <br /> [[nlp-progress 2016/11]]<br /> <br /> [[nlp-progress 2016/10]]<br /> <br /> [[nlp-progress 2016/09]]<br /> <br /> [[nlp-progress 2016/08]]<br /> <br /> [[nlp-progress 2016/05/01 -- 08/16 | nlp-progress 2016/05-07]]<br /> <br /> [[nlp-progress 2016/04]]</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/NLP_Status_Report_2016-12-19 2016-12-19T05:27:20Z

<p>Zhangandy：</p> <hr /> <div>{| class=&quot;wikitable&quot;<br /> !Date !! People !! Last Week !! This Week<br /> |-<br /> | rowspan=&quot;6&quot;|2016/12/19<br /> |Yang Feng ||<br /> *[[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]]<br /> *wrote part of the code of mn.<br /> *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]]<br /> *[[Huilan:]] fixed the problem of syntax-based translation.<br /> *sort out the system and corresponding documents.<br /> ||<br /> *[[s2smn:]] finish the code of adding mn.<br /> *[[Huilan:]] handover.<br /> |-<br /> |Jiyuan Zhang ||<br /> *attempted to use memory model to improve the atten model of bad effect<br /> *With the vernacular as the input，generated poem by local atten model[http://cslt.riit.tsinghua.edu.cn/mediawiki/images/2/2f/Local_atten_resluts.pdf]<br /> *Modified working mechanism of memory model(top1 to average)<br /> *help andi<br /> || <br /> *improve poem model <br /> |-<br /> |Andi Zhang ||<br /> *tried to modify the wrong softmax, but abandoned at last<br /> *added bleu scoring part<br /> ||<br /> *extract encoder outputs<br /> |-<br /> |Shiyue Zhang || <br /> * changed the one-hot vector to (0, -inf, -inf...), and retied the experiments. But no improvement showed.<br /> * tried 1-dim gate, but converged to baseline<br /> * tried to only train gate, but the best is taking all instance as &quot;right&quot;<br /> * trying a model similar to attention<br /> * [[http://cslt.riit.tsinghua.edu.cn/mediawiki/images/9/9f/RNNG%2Bmm_experiment_report.pdf report]]<br /> ||<br /> * try to add true action info when training gate<br /> * try different scale vectors<br /> * try to change cos to only inner product<br /> |-<br /> |Guli ||<br /> *read papers about Transfer learning and solving OOV<br /> *conducted comparative test <br /> *writing survey<br /> ||<br /> * complete the first draft of the survey <br /> |-<br /> |Peilun Xiao ||<br /> *Read a paper about document classification wiht GMM distributions of word vecotrs and try to code it in python<br /> *Use LDA to reduce the dimension of the text in r52、r8 and contrast the performance of classification<br /> ||<br /> *Use LDA to reduce the dimension of the text in 20news and webkb<br /> |}</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/Schedule 2016-12-18T14:55:02Z

<p>Zhangandy：/* Daily Report */</p> <hr /> <div>=NLP Schedule=<br /> <br /> ==Members==<br /> <br /> ===Current Members===<br /> <br /> * Yang Feng (冯洋)<br /> * Jiyuan Zhang （张记袁）<br /> * Aodong Li (李傲冬)<br /> * Andi Zhang (张安迪)<br /> * Shiyue Zhang (张诗悦)<br /> * Li Gu (古丽)<br /> * Peilun Xiao (肖培伦)<br /> <br /> ===Former Members===<br /> * '''Chao Xing (邢超)''' : FreeNeb<br /> * '''Rong Liu (刘荣)''' : 优酷<br /> * '''Xiaoxi Wang (王晓曦)''' : 图灵机器人<br /> * '''Xi Ma (马习)''' : 清华大学研究生<br /> * '''Tianyi Luo (骆天一)''' ： phd candidate in University of California Santa Cruz<br /> * '''Qixin Wang (王琪鑫)''' : MA candidate in University of California<br /> * '''DongXu Zhang (张东旭)''': --<br /> * '''Yiqiao Pan (潘一桥)''' ： MA candidate in University of Sydney <br /> * '''Shiyao Li （李诗瑶）''' : BUPT<br /> * '''Aiting Liu (刘艾婷)''' : BUPT<br /> <br /> ==Work Progress==<br /> ===Daily Report===<br /> <br /> {|class=&quot;wikitable&quot;<br /> ! Date !! Person !! start!! leave !! hours ||status<br /> |-<br /> | rowspan=&quot;3&quot;|2016/12/1<br /> |Andy Zhang|| 9:30||18:30 ||8 || <br /> *read source code<br /> *thought about ways to store encoder output<br /> |-<br /> |Shiyue Zhang || 9:10 || 19:00|| 8+ ||<br /> * draw tsne pictures<br /> |-<br /> |Guli ||9:10 ||9:50 ||12+ || <br /> *prepare data and run code<br /> *prepare for Thesis Report<br /> |- <br /> | rowspan=&quot;4&quot;|2016/12/2<br /> |Andy Zhang|| 9:30|| 18:30|| 8||<br /> *finish the code to output encoder output but have not tested yet<br /> *prepared a small data set for testing <br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||9:05 ||18:05 ||8 || <br /> *read code and check data <br /> *think about how to improve bleu score<br /> |- <br /> |Peilun Xiao ||10:00||12:30|| 2+||<br /> *learn basic things of work<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/5<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *group meeting<br /> *prepare the same data set as in Theano for Tensorflow<br /> |-<br /> |Shiyue Zhang || 9:30 || 19:00 || 8+||<br /> * discuss with teachers about the tsne pictures<br /> * group meeting<br /> |-<br /> |Guli ||9:00 ||22:35 ||13+ || <br /> *install moses<br /> *weekly report<br /> |- <br /> |Peilun Xiao ||9:30||18:30 ||8||<br /> *learn how to use weka<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/6<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *run fr-en (single encoder)<br /> *prepare paraphrase data set, enumerate them, excluding 呀啊哈, etc.<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||9:05 ||22;30 ||12+ || <br /> *run moses<br /> *write Thesis Report<br /> |- <br /> |Peilun Xiao ||9:40||18:30|| 8||<br /> *read a paper<br /> *use weka<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/7<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *prepare paraphrase data set, enumerate them, excluding 呀啊哈, etc.<br /> *met with some problems, used too much memory, had to restrict enumerate nums to 15000<br /> |-<br /> |Shiyue Zhang || 9:30 || 19:00 || 8+||<br /> * change mem to 28-dim states<br /> * retry previous experiments<br /> |-<br /> |Guli ||9:15||22;00||12+ || <br /> *run moses<br /> *write Thesis Report<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/8<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *finished preparing the data set<br /> *start working on bidirectional model <br /> |-<br /> |Shiyue Zhang || 9:30 || 19:00 || 8+||<br /> * try more experiments, but all are not good, always converge to baseline<br /> * draw the tsne picture of test data<br /> * discuss with Teacher Wang<br /> |-<br /> |Guli ||9:15||21;15||12 || <br /> *run moses<br /> *write Thesis Report<br /> |- <br /> |Peilun Xiao ||9:30||17:30|| 7||<br /> *learn GibbsLda++<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/9<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *finished coding, added masks for encoder inputs<br /> *had a some breakthrough with Jiyuan's help, but still meet nan problem <br /> |-<br /> |Shiyue Zhang ||9:30 || 19:00 || 8+||<br /> * return to 384-dim mem, which has 0.07 improvement than baseline<br /> * replace cos with a one-layer neural network, which is also slightly better than baseline<br /> * discuss with teachers, and find the problem of scale<br /> |-<br /> |Guli ||9:15||21;30||12 +|| <br /> *prepared Thesis Report<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/12<br /> |Andy Zhang|| || || ||<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||10:00||19;00|| 9|| <br /> *read paper to solve oov<br /> |- <br /> |Peilun Xiao ||9:30||19:00||9+||<br /> *use lda to generate document vector<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/13<br /> |Andy Zhang|| 9:30 || 18:30 || 8 ||<br /> *tried to modify the wrong softmax<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||09:30||21;00||11+|| <br /> *read papers<br /> |- <br /> |Peilun Xiao ||9:50 ||19:00 || 9 ||<br /> *use lda to generate document vector<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/14<br /> |Andy Zhang|| 9:30 || 18:30 || 8 ||<br /> *tried to modify the wrong softmax<br /> *decided to abandon it for now<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||09:20||20;30||11+|| <br /> *read papers<br /> *writing a survey<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/15<br /> |Andy Zhang|| 9:30 || 18:30 || 8 ||<br /> *tried to add bleu scoring into the code, but met with out of memory problem<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||09:20||20;30||11+|| <br /> *read papers<br /> *writing a survey<br /> |- <br /> |Peilun Xiao ||13:30||19:00||5+||<br /> *use lda to generate document vector<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/16<br /> |Andy Zhang|| 9:20 || 18:20 || 8 ||<br /> *finished adding bleu scoring into the code<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||09:30||20;30||11|| <br /> *Check the data, modify the error<br /> *writing a survey<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> |}<br /> <br /> ===Time Off Table===<br /> <br /> {| class=&quot;wikitable&quot;<br /> ! Date !! Yang Feng !! Jiyuan Zhang <br /> |-<br /> |}<br /> <br /> ==Past progress==<br /> <br /> [[nlp-progress 2016/11]]<br /> <br /> [[nlp-progress 2016/10]]<br /> <br /> [[nlp-progress 2016/09]]<br /> <br /> [[nlp-progress 2016/08]]<br /> <br /> [[nlp-progress 2016/05/01 -- 08/16 | nlp-progress 2016/05-07]]<br /> <br /> [[nlp-progress 2016/04]]</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/NLP_Status_Report_2016-12-12 2016-12-13T13:02:58Z

<p>Zhangandy：</p> <hr /> <div>{| class=&quot;wikitable&quot;<br /> !Date !! People !! Last Week !! This Week<br /> |-<br /> | rowspan=&quot;5&quot;|2016/12/05<br /> |Yang Feng ||<br /> *[[s2smn:]] installed tensorflow and ran a toy example (solved problems: version conflict and memory exhausted)<br /> *wrote the code of the memory network part<br /> *[[Huilan:]] prepared for periodical report and system submission.<br /> ||<br /> *[[s2smn:]] finish the code of mn<br /> *[[poemgen:]] experiment design<br /> *[[Huilan:]] system submission<br /> |-<br /> |Jiyuan Zhang ||<br /> *attempted to use memory model to improve the atten model of bad effect<br /> *With the vernacular as the input，generated poem by local atten model[http://cslt.riit.tsinghua.edu.cn/mediawiki/images/2/2f/Local_atten_resluts.pdf]<br /> *Modified working mechanism of memory model(top1 to average)<br /> *help andi<br /> || <br /> *improve poem model <br /> |-<br /> |Andi Zhang ||<br /> *prepared a paraphrase data set that is enumerated from a previous one (ignoring words like &quot;啊呀哈&quot;)<br /> *worked on coding bidirectional model under tensorflow, met with NAN problem<br /> ||<br /> *ignore NAN problem for now, run it on the same data set used in Theano<br /> |-<br /> |Shiyue Zhang || <br /> ||<br /> |-<br /> |Guli ||<br /> *install and run moses<br /> *prepare thesis report<br /> ||<br /> *read papers about Transfer learning and solving OOV <br /> |}</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/Schedule 2016-12-13T01:57:26Z

<p>Zhangandy：/* Daily Report */</p> <hr /> <div>=NLP Schedule=<br /> <br /> ==Members==<br /> <br /> ===Current Members===<br /> <br /> * Yang Feng (冯洋)<br /> * Jiyuan Zhang （张记袁）<br /> * Aodong Li (李傲冬)<br /> * Andi Zhang (张安迪)<br /> * Shiyue Zhang (张诗悦)<br /> * Li Gu (古丽)<br /> * Peilun Xiao (肖培伦)<br /> <br /> ===Former Members===<br /> * '''Chao Xing (邢超)''' : FreeNeb<br /> * '''Rong Liu (刘荣)''' : 优酷<br /> * '''Xiaoxi Wang (王晓曦)''' : 图灵机器人<br /> * '''Xi Ma (马习)''' : 清华大学研究生<br /> * '''Tianyi Luo (骆天一)''' ： phd candidate in University of California Santa Cruz<br /> * '''Qixin Wang (王琪鑫)''' : MA candidate in University of California<br /> * '''DongXu Zhang (张东旭)''': --<br /> * '''Yiqiao Pan (潘一桥)''' ： MA candidate in University of Sydney <br /> * '''Shiyao Li （李诗瑶）''' : BUPT<br /> * '''Aiting Liu (刘艾婷)''' : BUPT<br /> <br /> ==Work Progress==<br /> ===Daily Report===<br /> <br /> {|class=&quot;wikitable&quot;<br /> ! Date !! Person !! start!! leave !! hours ||status<br /> |-<br /> | rowspan=&quot;3&quot;|2016/12/1<br /> |Andy Zhang|| 9:30||18:30 ||8 || <br /> *read source code<br /> *thought about ways to store encoder output<br /> |-<br /> |Shiyue Zhang || 9:10 || 19:00|| 8+ ||<br /> * draw tsne pictures<br /> |-<br /> |Guli ||9:10 ||9:50 ||12+ || <br /> *prepare data and run code<br /> *prepare for Thesis Report<br /> |- <br /> | rowspan=&quot;4&quot;|2016/12/2<br /> |Andy Zhang|| 9:30|| 18:30|| 8||<br /> *finish the code to output encoder output but have not tested yet<br /> *prepared a small data set for testing <br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||9:05 ||18:05 ||8 || <br /> *read code and check data <br /> *think about how to improve bleu score<br /> |- <br /> |Peilun Xiao ||10:00||12:30|| 2+||<br /> *learn basic things of work<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/5<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *group meeting<br /> *prepare the same data set as in Theano for Tensorflow<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||9:00 ||22:35 ||13+ || <br /> *install moses<br /> *weekly report<br /> |- <br /> |Peilun Xiao ||9:30||18:30 ||8||<br /> *learn how to use weka<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/6<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *run fr-en (single encoder)<br /> *prepare paraphrase data set, enumerate them, excluding 呀啊哈, etc.<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||9:05 ||22;30 ||12+ || <br /> *run moses<br /> *write Thesis Report<br /> |- <br /> |Peilun Xiao ||9:40||18:30|| 8||<br /> *read a paper<br /> *use weka<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/7<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *prepare paraphrase data set, enumerate them, excluding 呀啊哈, etc.<br /> *met with some problems, used too much memory, had to restrict enumerate nums to 15000<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||9:15||22;00||12+ || <br /> *run moses<br /> *write Thesis Report<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/8<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *finished preparing the data set<br /> *start working on bidirectional model <br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||9:15||21;15||12 || <br /> *run moses<br /> *write Thesis Report<br /> |- <br /> |Peilun Xiao ||9:30||17:30|| 7||<br /> *learn GibbsLda++<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/9<br /> |Andy Zhang|| 9:30 || 19:00 || 8+ ||<br /> *finished coding, added masks for encoder inputs<br /> *had a some breakthrough with Jiyuan's help, but still meet nan problem <br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||9:15||21;30||12 +|| <br /> *prepared Thesis Report<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> | rowspan=&quot;4&quot;|2016/12/12<br /> |Andy Zhang|| || || ||<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||10:00||19;00|| 9|| <br /> *read paper to solve oov<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> |}<br /> <br /> ===Time Off Table===<br /> <br /> {| class=&quot;wikitable&quot;<br /> ! Date !! Yang Feng !! Jiyuan Zhang <br /> |-<br /> |}<br /> <br /> ==Past progress==<br /> <br /> [[nlp-progress 2016/11]]<br /> <br /> [[nlp-progress 2016/10]]<br /> <br /> [[nlp-progress 2016/09]]<br /> <br /> [[nlp-progress 2016/08]]<br /> <br /> [[nlp-progress 2016/05/01 -- 08/16 | nlp-progress 2016/05-07]]<br /> <br /> [[nlp-progress 2016/04]]</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/NLP_Status_Report_2016-12-05 2016-12-05T05:45:25Z

<p>Zhangandy：</p> <hr /> <div>{| class=&quot;wikitable&quot;<br /> !Date !! People !! Last Week !! This Week<br /> |-<br /> | rowspan=&quot;5&quot;|2016/12/05<br /> |Yang Feng ||<br /> *[[rnng+MN:]] got the result of k-means method and the result is slightly worse;<br /> *fixed the bug; <br /> *analyzed the memory units and changed the calculation of similarity and reran.<br /> *[[S2S+MN:]] read the code and discuss with andy about the implementation details; <br /> *checked Wikianswers data and found the answers are usually much longer than the question; <br /> *read 12 QA-related papers in proceedings of ACL16 and EMNLP16 and haven't found proper dataset yet.<br /> *[[Huilan's work:]] got a version of better result focusing on syntactical transformation.<br /> ||<br /> *[[rnng+MN:]] get the result with new similarity calculation.<br /> *[[S2S+MN:]] revise the code of tensorflow to make it equivalent to theano's. <br /> *[[poetry:]] review the code of Jiyuan<br /> *[[Huilan's work:]] continue the work of adding syntactic information. <br /> |-<br /> |Jiyuan Zhang ||<br /> *restructured code<br /> *found the cause of cost randomness<br /> *modified memory weight,ran expriment [http://cslt.riit.tsinghua.edu.cn/mediawiki/images/c/c9/Yanqing-weight%282.0%29.pdf 言情风格][http://cslt.riit.tsinghua.edu.cn/mediawiki/images/e/e4/Biansaishi-weight%282.0%29.pdf 边塞风格]<br /> *read a paper[http://cslt.riit.tsinghua.edu.cn/mediawiki/images/1/1e/1610.09889v1.pdf]<br /> *simply expain my code to Miss Feng<br /> *discussed with liantian about the way of using tensorflow to realize his idea<br /> || <br /> *improve poem model <br /> |-<br /> |Andi Zhang ||<br /> *turn to focus on tensorflow codes<br /> *finish code outputing encoder outputs but have some problems with format<br /> ||<br /> *finish the part mentioned left<br /> <br /> |-<br /> |Shiyue Zhang || <br /> * found a bug in my code and modified it.<br /> * tried memory with gate and found a big problem of memory.<br /> * reran previous models, the results are not better than baseline. [[http://cslt.riit.tsinghua.edu.cn/mediawiki/images/2/2f/RNNG%2Bmm%E5%AE%9E%E9%AA%8C%E6%8A%A5%E5%91%8A.pdf report]]<br /> * reran the original model setting same seed, and got exactly same result.<br /> * published a TRP [http://cslt.riit.tsinghua.edu.cn/mediawiki/index.php/Publication-trp]<br /> ||<br /> * try to solve the problem of mem<br /> |-<br /> |Guli ||<br /> * preparing more data set<br /> * working on add translate module to code <br /> * writing a survey about Chinese-uyghur MT <br /> ||<br /> * prepare more data<br /> * prepare for Thesis Report<br /> |}</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/Schedule 2016-12-05T05:30:30Z

<p>Zhangandy：/* Daily Report */</p> <hr /> <div>=NLP Schedule=<br /> <br /> ==Members==<br /> <br /> ===Current Members===<br /> <br /> * Yang Feng (冯洋)<br /> * Jiyuan Zhang （张记袁）<br /> * Andi Zhang (张安迪)<br /> * Shiyue Zhang (张诗悦)<br /> * Li Gu (古丽)<br /> * Peilun Xiao (肖培伦)<br /> <br /> ===Former Members===<br /> * '''Chao Xing (邢超)''' : FreeNeb<br /> * '''Rong Liu (刘荣)''' : 优酷<br /> * '''Xiaoxi Wang (王晓曦)''' : 图灵机器人<br /> * '''Xi Ma (马习)''' : 清华大学研究生<br /> * '''Tianyi Luo (骆天一)''' ： phd candidate in University of California Santa Cruz<br /> * '''Qixin Wang (王琪鑫)''' : MA candidate in University of California<br /> * '''DongXu Zhang (张东旭)''': --<br /> * '''Yiqiao Pan (潘一桥)''' ： MA candidate in University of Sydney <br /> * '''Shiyao Li （李诗瑶）''' : BUPT<br /> * '''Aiting Liu (刘艾婷)''' : BUPT<br /> * '''Aodong Li (李傲冬)'''：BUPT<br /> <br /> ==Work Progress==<br /> ===Daily Report===<br /> <br /> {|class=&quot;wikitable&quot;<br /> ! Date !! Person !! start!! leave !! hours ||status<br /> |-<br /> | rowspan=&quot;3&quot;|2016/12/1<br /> |Andy Zhang|| 9:30||18:30 ||8 || <br /> *read source code<br /> *thought about ways to store encoder output<br /> |-<br /> |Shiyue Zhang || 9:10 || 19:00|| 8+ ||<br /> * draw tsne pictures<br /> |-<br /> |Guli ||9:10 ||9:50 ||12+ || <br /> *prepare data and run code<br /> *prepare for Thesis Report<br /> |- <br /> | rowspan=&quot;4&quot;|2016/12/2<br /> |Andy Zhang|| 9:30|| 18:30|| 8||<br /> *finish the code to output encoder output but have not tested yet<br /> *prepared a small data set for testing <br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> |Guli ||9:05 ||18:05 ||8 || <br /> *read code and check data <br /> *think about how to improve bleu score<br /> |- <br /> |Peilun Xiao || || || ||<br /> |-<br /> |}<br /> <br /> ===Time Off Table===<br /> <br /> {| class=&quot;wikitable&quot;<br /> ! Date !! Yang Feng !! Jiyuan Zhang <br /> |-<br /> |}<br /> <br /> ==Past progress==<br /> <br /> [[nlp-progress 2016/11]]<br /> <br /> [[nlp-progress 2016/10]]<br /> <br /> [[nlp-progress 2016/09]]<br /> <br /> [[nlp-progress 2016/08]]<br /> <br /> [[nlp-progress 2016/05/01 -- 08/16 | nlp-progress 2016/05-07]]<br /> <br /> [[nlp-progress 2016/04]]</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/Cslt-member-visitors 2016-12-05T00:48:25Z

<p>Zhangandy：/* Andy Zhang(张安迪) */</p> <hr /> <div><br /> ==Professionals==<br /> <br /> === Yang Feng (冯洋) ===<br /> [[文件:Fengyang.jpg| 200px]]<br /> * 2016.6.21-<br /> * Huilan Ltd<br /> * language processing<br /> <br /> <br /> ==Engineers==<br /> <br /> === Yuxin Zhang (张雨心） ===<br /> [[文件:Daisy.jpg|200px]]<br /> * Haixia research center<br /> * 2016.10 -<br /> * Finance processing<br /> <br /> <br /> ==Students==<br /> <br /> <br /> ===Jiyuan Zhang (张记袁）===<br /> [[文件:Zhangjiyuan.png|200px]]<br /> * PKU<br /> * 2016.4-<br /> * neural generation model<br /> * [[媒体文件:An overview of machine translation.pptx|Bi-weekly report]]<br /> * [[媒体文件:Zhangjy_data.jpg|Data_security_agreement]]<br /> <br /> ===Jingyi Lin(林婧伊)===<br /> [[文件:JinyiLin_photo.jpg|200px]]<br /> * China chemical university<br /> * 2016.4.12-<br /> * Speech processing<br /> * [[媒体文件:Bi_weekly_report_linjy.ppt|Bi-weekly report]]<br /> *[[媒体文件:DataSecurityAgreement_JingyiLin.jpg|DataSecurityAgreement]]<br /> <br /> ===Ziwei Bai (白子薇）===<br /> [[文件:Ziwei Bai.jpg|200px]]<br /> * BUPT<br /> * 2016.5-<br /> * QA<br /> * [[媒体文件:xxx|Bi-weekly report]]<br /> <br /> ===Ying Shi (石颖）===<br /> [[文件:Ying_shi.jpg|200px]]<br /> * BJTU<br /> * 2016.6.15-<br /> * Speech processing<br /> * [[媒体文件:Shiying_bi_weekly_report.ppt|Bi-weekly report]]<br /> *[[媒体文件:DataSecurityAgreement_YingShi.jpg|DataSecurityAgreement]]<br /> <br /> ===Aodong Li (李傲冬）===<br /> [[文件:李傲冬.jpg|200px]]<br /> * BUPT<br /> * 2016.6-<br /> * Neural generation model<br /> * [[媒体文件:Biweekly_report_aodong.pptx|Bi-weekly report]]<br /> <br /> ===Yixiang Chen (陈怿翔)===<br /> [[文件:Chenyx.jpg|200px]]<br /> *University of China Mining<br /> * 2016.7-<br /> * Speech processing<br /> * [[媒体文件:Chenyx_report.pdf |Bi-weekly report]]<br /> * [[媒体文件:Chenyx data.jpg|Data_security_agreement]]<br /> <br /> ===Shiyue Zhang(张诗悦)===<br /> [[文件:Zhang Shiyue.jpg|200px]]<br /> * BUTP<br /> * 2016.9.06-<br /> * Language processing<br /> * [[媒体文件:1.pic hd.jpg| Data_security_agreement]]<br /> <br /> ===Andy Zhang(张安迪)===<br /> [[文件:andyzhang.jpg|200px]]<br /> * BUPT<br /> * 2016.07.19-<br /> * Language processing<br /> * [[媒体文件:AndyZhang汇报.pptx|Bi-weekly report]]<br /> * [[媒体文件:Data agreement andy.jpg|Data_security_agreement]]<br /> <br /> ===Hang Luo (罗航)===<br /> [[文件:罗航.png|200px]]<br /> * PKU<br /> * 2016.10-<br /> * Speech processing<br /> * [[媒体文件:Bi-weekly_work_report_Luohang.pptx|Bi-weekly report]]<br /> * [[媒体文件:Security.jpg|Data_security_agreement]]<br /> <br /> ===Bin Fu (符宾）===<br /> [[文件:Fubin.jpg|200px]]<br /> * Dalian Haishi<br /> * 2016.10-<br /> * Speech processing<br /> <br /> ===Li Gu (古丽)===<br /> [[文件:guli.jpg|200px]]<br /> * Xingjiang U.<br /> * 2016.11-<br /> * Translation<br /> *[[媒体文件:3.jpg|Data_security_agreement]]<br /> <br /> ===Yanqing Wang（王延清）===<br /> [[文件:wyq photo.jpeg|200px]]<br /> * BUPT<br /> * 2016.11-<br /> * Signal processing<br /> * [[媒体文件:Bi-weekly_report.pptx|Bi-weekly report]]<br /> *[[媒体文件:DataSecurityAgreement wangyanqing.jpg|Data_security_agreement]]<br /> <br /> ===Peilun Xiao（肖培伦）===<br /> [[文件:Daisy.jpg|200px]]<br /> * BUPT<br /> * 2016.12-<br /> * NLP<br /> * [[媒体文件:xxx|Bi-weekly report]]<br /> <br /> ==Intern==<br /> <br /> === Yang Wei (魏扬） ===<br /> [[文件:weiy_photo.jpg|200px]]<br /> * 2016.10 -<br /> * Speech processing</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/%E6%96%87%E4%BB%B6:Data_agreement_andy.jpg 2016-12-05T00:47:14Z

<p>Zhangandy：</p> <hr /> <div></div>

Zhangandy http://index.cslt.org/mediawiki/index.php/Cslt-member-visitors 2016-12-03T07:02:35Z

<p>Zhangandy：/* Andy Zhang(张安迪) */</p> <hr /> <div><br /> ==Professionals==<br /> <br /> ===Maoning Wang (王卯宁）===<br /> [[文件:Daisy.jpg|200px]]<br /> * Centeral Fiance and Economic University<br /> * 2016.9-<br /> * Finance processing<br /> <br /> <br /> ===Caixia Wang (王彩霞）===<br /> [[文件:Daisy.jpg|200px]]<br /> * University of diplomatic<br /> * 2016.9-<br /> * Finance processing<br /> <br /> <br /> <br /> === Yang Feng (冯洋) ===<br /> [[文件:Fengyang.jpg| 200px]]<br /> * 2016.6.21-<br /> * Huilan Ltd<br /> * language processing<br /> <br /> <br /> ==Engineers==<br /> <br /> === Yuxin Zhang (张雨心） ===<br /> [[文件:Daisy.jpg|200px]]<br /> * Haixia research center<br /> * 2016.10 -<br /> * Finance processing<br /> <br /> <br /> ==Students==<br /> <br /> <br /> ===Jiyuan Zhang (张记袁）===<br /> [[文件:Zhangjiyuan.png|200px]]<br /> * PKU<br /> * 2016.4-<br /> * neural generation model<br /> * [[媒体文件:xxx|Bi-weekly report]]<br /> <br /> ===Jingyi Lin(林婧伊)===<br /> [[文件:JinyiLin_photo.jpg|200px]]<br /> * China chemical university<br /> * 2016.4.12-<br /> * Speech processing<br /> * [[媒体文件:Bi_weekly_report_linjy.ppt|Bi-weekly report]]<br /> <br /> ===Ziwei Bai (白子薇）===<br /> [[文件:Ziwei Bai.jpg|200px]]<br /> * BUPT<br /> * 2016.5-<br /> * QA<br /> * [[媒体文件:xxx|Bi-weekly report]]<br /> <br /> ===Ying Shi (石颖）===<br /> [[文件:Ying_shi.jpg|200px]]<br /> * BJTU<br /> * 2016.6.15-<br /> * Speech processing<br /> * [[媒体文件:Shiying_bi_weekly_report.ppt|Bi-weekly report]]<br /> <br /> ===Aodong Li (李傲冬）===<br /> [[文件:Daisy.jpg|200px]]<br /> * BUPT<br /> * 2016.6-<br /> * Neural generation model<br /> * [[媒体文件:xxx|Bi-weekly report]]<br /> <br /> ===Yixiang Chen (陈怿翔)===<br /> [[文件:Chenyx.jpg|200px]]<br /> *University of China Mining<br /> * 2016.7-<br /> * Speech processing<br /> * [[媒体文件:xxx|Bi-weekly report]]<br /> <br /> ===Shiyue Zhang(张诗悦)===<br /> [[文件:Zhang Shiyue.jpg|200px]]<br /> * BUTP<br /> * 2016.9.06-<br /> * Language processing<br /> * [[媒体文件:xxx|Bi-weekly report]]<br /> <br /> ===Andy Zhang(张安迪)===<br /> [[文件:andyzhang.jpg|200px]]<br /> * BUPT<br /> * 2016.07.19-<br /> * Language processing<br /> * [[媒体文件:AndyZhang汇报.pptx|Bi-weekly report]]<br /> <br /> ===Hang Luo (罗航)===<br /> [[文件:罗航.png|200px]]<br /> * PKU<br /> * 2016.10-<br /> * Speech processing<br /> * [[媒体文件:Bi-weekly_work_report_Luohang.pptx|Bi-weekly report]]<br /> <br /> ===Bin Fu (符宾）===<br /> [[文件:Fubin.jpg|200px]]<br /> * Dalian Haishi<br /> * 2016.10-<br /> * Speech processing<br /> <br /> ===Li Gu (古丽)===<br /> [[文件:Daisy.jpg|200px]]<br /> * Xingjiang U.<br /> * 2016.11-<br /> * Translation<br /> <br /> <br /> ===Yanqing Wang（王延清）===<br /> [[文件:wyq photo.jpeg|200px]]<br /> * BUPT<br /> * 2016.11-<br /> * Signal processing<br /> * [[媒体文件:Bi-weekly_report.pptx|Bi-weekly report]]<br /> <br /> ===Peilun Xiao（肖培伦）===<br /> [[文件:Daisy.jpg|200px]]<br /> * BUPT<br /> * 2016.12-<br /> * NLP<br /> * [[媒体文件:xxx|Bi-weekly report]]<br /> <br /> ==Intern==<br /> <br /> === Yang Wei (魏扬） ===<br /> [[文件:weiy_photo.jpg|200px]]<br /> * 2016.10 -<br /> * Speech processing</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/%E6%96%87%E4%BB%B6:Andyzhang.jpg 2016-12-03T07:02:13Z

<p>Zhangandy：</p> <hr /> <div></div>

Zhangandy http://index.cslt.org/mediawiki/index.php/Schedule 2016-12-02T01:56:36Z

<p>Zhangandy：/* Daily Report */</p> <hr /> <div>=NLP Schedule=<br /> <br /> ==Members==<br /> <br /> ===Current Members===<br /> <br /> * Yang Feng (冯洋)<br /> * Jiyuan Zhang （张记袁）<br /> * Andi Zhang (张安迪)<br /> * Shiyue Zhang (张诗悦)<br /> * Li Gu (古丽)<br /> * Peilun Xiao (肖培伦)<br /> <br /> ===Former Members===<br /> * '''Chao Xing (邢超)''' : FreeNeb<br /> * '''Rong Liu (刘荣)''' : 优酷<br /> * '''Xiaoxi Wang (王晓曦)''' : 图灵机器人<br /> * '''Xi Ma (马习)''' : 清华大学研究生<br /> * '''Tianyi Luo (骆天一)''' ： phd candidate in University of California Santa Cruz<br /> * '''Qixin Wang (王琪鑫)''' : MA candidate in University of California<br /> * '''DongXu Zhang (张东旭)''': --<br /> * '''Yiqiao Pan (潘一桥)''' ： MA candidate in University of Sydney <br /> * '''Shiyao Li （李诗瑶）''' : BUPT<br /> * '''Aiting Liu (刘艾婷)''' : BUPT<br /> * '''Aodong Li (李傲冬)'''：BUPT<br /> <br /> ==Work Progress==<br /> ===Daily Report===<br /> <br /> {|class=&quot;wikitable&quot;<br /> ! Date !! Person !! start!! leave !! hours ||status<br /> |-<br /> | rowspan=&quot;3&quot;|2016/12/1<br /> |Andy Zhang|| 9:30||18:30 ||8 || <br /> *read source code<br /> *thought about ways to store encoder output<br /> |-<br /> |Shiyue Zhang || || || ||<br /> <br /> |-<br /> |Guli || || || || <br /> |-<br /> Peilun Xiao || || || ||<br /> |}<br /> <br /> ===Time Off Table===<br /> <br /> {| class=&quot;wikitable&quot;<br /> ! Date !! Yang Feng !! Jiyuan Zhang <br /> |-<br /> |}<br /> <br /> ==Past progress==<br /> <br /> [[nlp-progress 2016/11]]<br /> <br /> [[nlp-progress 2016/10]]<br /> <br /> [[nlp-progress 2016/09]]<br /> <br /> [[nlp-progress 2016/08]]<br /> <br /> [[nlp-progress 2016/05/01 -- 08/16 | nlp-progress 2016/05-07]]<br /> <br /> [[nlp-progress 2016/04]]</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/Schedule 2016-11-30T14:43:20Z

<p>Zhangandy：/* Daily Report */</p> <hr /> <div>=NLP Schedule=<br /> <br /> ==Members==<br /> <br /> ===Current Members===<br /> <br /> * Yang Feng (冯洋)<br /> * Jiyuan Zhang （张记袁）<br /> * Aodong Li (李傲冬)<br /> * Andi Zhang (张安迪)<br /> * Aiting Liu (刘艾婷)<br /> * Shiyao Li （李诗瑶）<br /> * Shiyue Zhang (张诗悦)<br /> <br /> ===Former Members===<br /> * '''Chao Xing (邢超)''' : FreeNeb<br /> * '''Rong Liu (刘荣)''' : 优酷<br /> * '''Xiaoxi Wang (王晓曦)''' : 图灵机器人<br /> * '''Xi Ma (马习)''' : 清华大学研究生<br /> * '''Tianyi Luo (骆天一)''' ： phd candidate in University of California Santa Cruz<br /> * '''Qixin Wang (王琪鑫)''' : MA candidate in University of California<br /> * '''DongXu Zhang (张东旭)''': --<br /> * '''Yiqiao Pan (潘一桥)''' ： MA candidate in University of Sydney <br /> <br /> <br /> ==Work Progress==<br /> ===Daily Report===<br /> <br /> {|class=&quot;wikitable&quot;<br /> ! Date !! Person !! start!! leave !! hours ||status<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/1<br /> |Andy Zhang|| 10:00|| 19:00||8 || <br /> *read source code<br /> *weekly meeting <br /> |-<br /> |Shiyue Zhang || 9:15 || 19:30|| 8||<br /> * finish a simple way to add memory into rnng and write a report<br /> * meeting <br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/2<br /> |Andy Zhang|| 10:00|| 19:00||8 ||<br /> *tried to figure out the cost type of seq2seq<br /> *bimonthly report <br /> |-<br /> |Shiyue Zhang || 9:15 || 20:00|| 8+||<br /> * fix unexpected action<br /> * rerun original model<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/3<br /> |Andy Zhang|| 13:30|| 20:20|| 6|| <br /> *read source code <br /> |-<br /> |Shiyue Zhang || 9:35 || 19:00|| 8+||<br /> * rerun center memory model<br /> * run sample memory model<br /> * run wrong memory model<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/4<br /> |Andy Zhang|| 10:00|| 18:30|| 7+|| <br /> *try to run code on GPU but failed <br /> |-<br /> |Shiyue Zhang || |||| ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/7<br /> |Andy Zhang|| 10:00|| 19:20|| 8+|| <br /> *try to run code on GPU but failed <br /> |-<br /> |Shiyue Zhang || 9:15 || 20:30|| 9 ||<br /> * review last week work<br /> * write report <br /> * meeting<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/8<br /> |Andy Zhang|| 9:40|| 19:00|| 8+|| <br /> *finally ran the code on gpu, waiting for results to continue<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00 || 8+||<br /> * try to run rnng on GPU, but fail<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/9<br /> |Andy Zhang||10:00 ||19:00 ||8 || <br /> *bugs when doing validation, solved it but had to rerun the code<br /> *writing document of MemN2N<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00 || 8+||<br /> * try several wrong mem models<br /> * rebuild rnng on Dynet<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/10<br /> |Andy Zhang|| 9:30||18:30 || 8|| <br /> <br /> *writing document of MemN2N<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00|| 9||<br /> * read the code of Feng<br /> * try to run rnng on GPU, but fail<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/11<br /> |Andy Zhang|| 9:30||18:30 || 8|| <br /> *ran NTM on paraphrase data set<br /> <br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/14<br /> |Andy Zhang||9:30 ||18:30|| 8|| <br /> *prepared data set of fr-en for NMT; prepare training, val &amp; test set for paraphrase<br /> *ran NMT model on the above data<br /> *weekly meeting<br /> |-<br /> |Shiyue Zhang || 9:30 || 21:30|| 9+||<br /> * review last week work<br /> * meeting<br /> * try MKL, but it cannot use multi cpu cores<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/15<br /> |Andy Zhang||9:30 ||18:30|| 8|| <br /> *deal with paraphrase data set<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00 || 8+||<br /> * run rnng on MKL successfully, which can at least double the speed.<br /> * run the original rnng discriminative model<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/16<br /> |Andy Zhang||10:00 ||19:00|| 8|| <br /> *deal with paraphrase data set, wipe out repetitions and some noises<br /> *help Guli with NMT code<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00 || 8+||<br /> * finish the code of dynamic memory model and running<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/17<br /> |Andy Zhang||10:00 ||19:00||8 || <br /> *run NTM on data dealt yesterday<br /> *read through source code to find ways to modify it<br /> |-<br /> |Shiyue Zhang || 9:30 || 17:00|| 6 ||<br /> * try the memory structure mentioned by Teacher Wang<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/18<br /> |Andy Zhang||10:00 ||19:00||8 || <br /> *read through source code to find ways to modify it<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/21<br /> |Andy Zhang||9:30 ||19:00||8+ || <br /> *read source code and seem to figure out how to modify it<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00 || 8+||<br /> * review the work of last week and meeting<br /> * rerun original model, try more dynamic models<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/22<br /> |Andy Zhang||9:30 ||19:00||8+ || <br /> *deal with zh2en data set<br /> *run NTM on them<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/23<br /> |Andy Zhang||9:30 ||19:00||8+ || <br /> *discuss code with Mrs. Feng<br /> |-<br /> |Shiyue Zhang || 9:10 || 6:30 || 8+||<br /> * find a big bug in my code and modify it<br /> * try the second memory structure with gate, and find a problem of memory<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/24<br /> |Andy Zhang||10:00 ||19:00||8 || <br /> *met with crushes <br /> *figure out relationship between checkpoint model &amp; best model<br /> |-<br /> |Shiyue Zhang || 9:10 || 6:30 || 8+||<br /> * discuss the problem with Teacher Feng and Wang<br /> * think about the solution of the problem<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/25<br /> |Andy Zhang||10:00||14:00||3+|| <br /> *think about ways to output the result of encoder<br /> *asked for a leave<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/28<br /> |Andy Zhang||9:30||18:30||8+|| <br /> *find ways to output the result of encoder<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/29<br /> |Andy Zhang||9:30||18:30||8+|| <br /> *turn to tensorflow<br /> *read source code of seq2seq of tensorflow<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/30<br /> |Andy Zhang||9:30||18:30||8+|| <br /> *read source code of seq2seq of tensorflow<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |}<br /> <br /> ===Monthly Summary===<br /> <br /> {| class=&quot;wikitable&quot;<br /> !People!! Summary<br /> |-<br /> |Yang Feng || <br /> |-<br /> |Jiyuan Zhang || <br /> |-<br /> |Andy Zhang || <br /> |-<br /> |Shiyue Zhang ||<br /> |}<br /> <br /> ===Time Off Table===<br /> <br /> {| class=&quot;wikitable&quot;<br /> ! Date !! Yang Feng !! Jiyuan Zhang <br /> |-<br /> |11/01 || 8h || <br /> |-<br /> |11/21 || 2h ||<br /> |-<br /> |11/28 || 8h ||<br /> |-<br /> |11/29 || 2h ||<br /> |}<br /> <br /> ==Past progress==<br /> [[nlp-progress 2016/10]]<br /> <br /> [[nlp-progress 2016/09]]<br /> <br /> [[nlp-progress 2016/08]]<br /> <br /> [[nlp-progress 2016/05/01 -- 08/16 | nlp-progress 2016/05-07]]<br /> <br /> [[nlp-progress 2016/04]]</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/NLP_Status_Report_2016-11-28 2016-11-29T04:42:31Z

<p>Zhangandy：</p> <hr /> <div>{| class=&quot;wikitable&quot;<br /> !Date !! People !! Last Week !! This Week<br /> |-<br /> | rowspan=&quot;5&quot;|2016/11/28<br /> |Yang Feng ||<br /> ||<br /> |-<br /> |Jiyuan Zhang ||<br /> *read three related paper to find some ideas &lt;br/&gt; [http://cslt.riit.tsinghua.edu.cn/mediawiki/images/3/3e/1410.5401v2.pdf neural turing machine]&lt;br/&gt;[http://cslt.riit.tsinghua.edu.cn/mediawiki/images/5/54/1508.06576v1.pdf A Neural Algorithm of Artistic Style]&lt;br/&gt;[http://cslt.riit.tsinghua.edu.cn/mediawiki/images/8/83/Dedaff23ad393c48fe7b7989542318a02dc0a06e.pdf Generating Long and Diverse Responses with Neural Conversation Models]<br /> <br /> * polished TRP [http://cslt.riit.tsinghua.edu.cn/mediawiki/images/c/c4/Memory-atten-model-public.pdf]<br /> || <br /> *improve poem model <br /> |-<br /> |Andi Zhang ||<br /> *deal with zh2en data set and ran them on NTM<br /> *had a small breakthrough about the code<br /> <br /> ||<br /> *get output of encoder to form memory<br /> *continue on the coding work of seq2seq with MemN2N<br /> |-<br /> |Shiyue Zhang || <br /> * found a bug in my code and modified it.<br /> * tried memory with gate and found a big problem of memory.<br /> * reran previous models, the results are not better than baseline. [[http://cslt.riit.tsinghua.edu.cn/mediawiki/images/2/2f/RNNG%2Bmm%E5%AE%9E%E9%AA%8C%E6%8A%A5%E5%91%8A.pdf report]]<br /> * reran the original model setting same seed, and got exactly same result.<br /> * published a TRP [http://cslt.riit.tsinghua.edu.cn/mediawiki/index.php/Publication-trp]<br /> ||<br /> * try to solve the problem of mem<br /> |-<br /> |Guli ||<br /> * busy on nothing for the first two days of the week.<br /> * modify the code and run NMT on fr-en data set<br /> * modify the code and run NMT on ch-uy data set<br /> * writing a survey about Chinese-uyghur MT<br /> ||<br /> |}</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/Schedule 2016-11-29T04:33:33Z

<p>Zhangandy：/* Daily Report */</p> <hr /> <div>=NLP Schedule=<br /> <br /> ==Members==<br /> <br /> ===Current Members===<br /> <br /> * Yang Feng (冯洋)<br /> * Jiyuan Zhang （张记袁）<br /> * Aodong Li (李傲冬)<br /> * Andi Zhang (张安迪)<br /> * Aiting Liu (刘艾婷)<br /> * Shiyao Li （李诗瑶）<br /> * Shiyue Zhang (张诗悦)<br /> <br /> ===Former Members===<br /> * '''Chao Xing (邢超)''' : FreeNeb<br /> * '''Rong Liu (刘荣)''' : 优酷<br /> * '''Xiaoxi Wang (王晓曦)''' : 图灵机器人<br /> * '''Xi Ma (马习)''' : 清华大学研究生<br /> * '''Tianyi Luo (骆天一)''' ： phd candidate in University of California Santa Cruz<br /> * '''Qixin Wang (王琪鑫)''' : MA candidate in University of California<br /> * '''DongXu Zhang (张东旭)''': --<br /> * '''Yiqiao Pan (潘一桥)''' ： MA candidate in University of Sydney <br /> <br /> <br /> ==Work Progress==<br /> ===Daily Report===<br /> <br /> {|class=&quot;wikitable&quot;<br /> ! Date !! Person !! start!! leave !! hours ||status<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/1<br /> |Andy Zhang|| 10:00|| 19:00||8 || <br /> *read source code<br /> *weekly meeting <br /> |-<br /> |Shiyue Zhang || 9:15 || 19:30|| 8||<br /> * finish a simple way to add memory into rnng and write a report<br /> * meeting <br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/2<br /> |Andy Zhang|| 10:00|| 19:00||8 ||<br /> *tried to figure out the cost type of seq2seq<br /> *bimonthly report <br /> |-<br /> |Shiyue Zhang || 9:15 || 20:00|| 8+||<br /> * fix unexpected action<br /> * rerun original model<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/3<br /> |Andy Zhang|| 13:30|| 20:20|| 6|| <br /> *read source code <br /> |-<br /> |Shiyue Zhang || 9:35 || 19:00|| 8+||<br /> * rerun center memory model<br /> * run sample memory model<br /> * run wrong memory model<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/4<br /> |Andy Zhang|| 10:00|| 18:30|| 7+|| <br /> *try to run code on GPU but failed <br /> |-<br /> |Shiyue Zhang || |||| ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/7<br /> |Andy Zhang|| 10:00|| 19:20|| 8+|| <br /> *try to run code on GPU but failed <br /> |-<br /> |Shiyue Zhang || 9:15 || 20:30|| 9 ||<br /> * review last week work<br /> * write report <br /> * meeting<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/8<br /> |Andy Zhang|| 9:40|| 19:00|| 8+|| <br /> *finally ran the code on gpu, waiting for results to continue<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00 || 8+||<br /> * try to run rnng on GPU, but fail<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/9<br /> |Andy Zhang||10:00 ||19:00 ||8 || <br /> *bugs when doing validation, solved it but had to rerun the code<br /> *writing document of MemN2N<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00 || 8+||<br /> * try several wrong mem models<br /> * rebuild rnng on Dynet<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/10<br /> |Andy Zhang|| 9:30||18:30 || 8|| <br /> <br /> *writing document of MemN2N<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00|| 9||<br /> * read the code of Feng<br /> * try to run rnng on GPU, but fail<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/11<br /> |Andy Zhang|| 9:30||18:30 || 8|| <br /> *ran NTM on paraphrase data set<br /> <br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/14<br /> |Andy Zhang||9:30 ||18:30|| 8|| <br /> *prepared data set of fr-en for NMT; prepare training, val &amp; test set for paraphrase<br /> *ran NMT model on the above data<br /> *weekly meeting<br /> |-<br /> |Shiyue Zhang || 9:30 || 21:30|| 9+||<br /> * review last week work<br /> * meeting<br /> * try MKL, but it cannot use multi cpu cores<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/15<br /> |Andy Zhang||9:30 ||18:30|| 8|| <br /> *deal with paraphrase data set<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00 || 8+||<br /> * run rnng on MKL successfully, which can at least double the speed.<br /> * run the original rnng discriminative model<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/16<br /> |Andy Zhang||10:00 ||19:00|| 8|| <br /> *deal with paraphrase data set, wipe out repetitions and some noises<br /> *help Guli with NMT code<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00 || 8+||<br /> * finish the code of dynamic memory model and running<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/17<br /> |Andy Zhang||10:00 ||19:00||8 || <br /> *run NTM on data dealt yesterday<br /> *read through source code to find ways to modify it<br /> |-<br /> |Shiyue Zhang || 9:30 || 17:00|| 6 ||<br /> * try the memory structure mentioned by Teacher Wang<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/18<br /> |Andy Zhang||10:00 ||19:00||8 || <br /> *read through source code to find ways to modify it<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/21<br /> |Andy Zhang||9:30 ||19:00||8+ || <br /> *read source code and seem to figure out how to modify it<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00 || 8+||<br /> * review the work of last week and meeting<br /> * rerun original model, try more dynamic models<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/22<br /> |Andy Zhang||9:30 ||19:00||8+ || <br /> *deal with zh2en data set<br /> *run NTM on them<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/23<br /> |Andy Zhang||9:30 ||19:00||8+ || <br /> *discuss code with Mrs. Feng<br /> |-<br /> |Shiyue Zhang || 9:10 || 6:30 || 8+||<br /> * find a big bug in my code and modify it<br /> * try the second memory structure with gate, and find a problem of memory<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/24<br /> |Andy Zhang||10:00 ||19:00||8 || <br /> *met with crushes <br /> *figure out relationship between checkpoint model &amp; best model<br /> |-<br /> |Shiyue Zhang || 9:10 || 6:30 || 8+||<br /> * discuss the problem with Teacher Feng and Wang<br /> * think about the solution of the problem<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/25<br /> |Andy Zhang||10:00||14:00||3+|| <br /> *think about ways to output the result of encoder<br /> *asked for a leave<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |}<br /> <br /> ===Monthly Summary===<br /> <br /> {| class=&quot;wikitable&quot;<br /> !People!! Summary<br /> |-<br /> |Yang Feng || <br /> |-<br /> |Jiyuan Zhang || <br /> |-<br /> |Andy Zhang || <br /> |-<br /> |Shiyue Zhang ||<br /> |}<br /> <br /> ===Time Off Table===<br /> <br /> {| class=&quot;wikitable&quot;<br /> ! Date !! Yang Feng !! Jiyuan Zhang <br /> |-<br /> |11/01 || 8h || <br /> |-<br /> |11/21 || 2h ||<br /> |-<br /> |11/28 || 8h ||<br /> |-<br /> |11/29 || 2h ||<br /> |}<br /> <br /> ==Past progress==<br /> [[nlp-progress 2016/10]]<br /> <br /> [[nlp-progress 2016/09]]<br /> <br /> [[nlp-progress 2016/08]]<br /> <br /> [[nlp-progress 2016/05/01 -- 08/16 | nlp-progress 2016/05-07]]<br /> <br /> [[nlp-progress 2016/04]]</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/Schedule 2016-11-24T13:04:03Z

<p>Zhangandy：/* Daily Report */</p> <hr /> <div>=NLP Schedule=<br /> <br /> ==Members==<br /> <br /> ===Current Members===<br /> <br /> * Yang Feng (冯洋)<br /> * Jiyuan Zhang （张记袁）<br /> * Aodong Li (李傲冬)<br /> * Andi Zhang (张安迪)<br /> * Aiting Liu (刘艾婷)<br /> * Shiyao Li （李诗瑶）<br /> * Shiyue Zhang (张诗悦)<br /> <br /> ===Former Members===<br /> * '''Chao Xing (邢超)''' : FreeNeb<br /> * '''Rong Liu (刘荣)''' : 优酷<br /> * '''Xiaoxi Wang (王晓曦)''' : 图灵机器人<br /> * '''Xi Ma (马习)''' : 清华大学研究生<br /> * '''Tianyi Luo (骆天一)''' ： phd candidate in University of California Santa Cruz<br /> * '''Qixin Wang (王琪鑫)''' : MA candidate in University of California<br /> * '''DongXu Zhang (张东旭)''': --<br /> * '''Yiqiao Pan (潘一桥)''' ： MA candidate in University of Sydney <br /> <br /> <br /> ==Work Progress==<br /> ===Daily Report===<br /> <br /> {|class=&quot;wikitable&quot;<br /> ! Date !! Person !! start!! leave !! hours ||status<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/1<br /> |Andy Zhang|| 10:00|| 19:00||8 || <br /> *read source code<br /> *weekly meeting <br /> |-<br /> |Shiyue Zhang || 9:15 || 19:30|| 8||<br /> * finish a simple way to add memory into rnng and write a report<br /> * meeting <br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/2<br /> |Andy Zhang|| 10:00|| 19:00||8 ||<br /> *tried to figure out the cost type of seq2seq<br /> *bimonthly report <br /> |-<br /> |Shiyue Zhang || 9:15 || 20:00|| 8+||<br /> * fix unexpected action<br /> * rerun original model<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/3<br /> |Andy Zhang|| 13:30|| 20:20|| 6|| <br /> *read source code <br /> |-<br /> |Shiyue Zhang || 9:35 || 19:00|| 8+||<br /> * rerun center memory model<br /> * run sample memory model<br /> * run wrong memory model<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/4<br /> |Andy Zhang|| 10:00|| 18:30|| 7+|| <br /> *try to run code on GPU but failed <br /> |-<br /> |Shiyue Zhang || |||| ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/7<br /> |Andy Zhang|| 10:00|| 19:20|| 8+|| <br /> *try to run code on GPU but failed <br /> |-<br /> |Shiyue Zhang || 9:15 || 20:30|| 9 ||<br /> * review last week work<br /> * write report <br /> * meeting<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/8<br /> |Andy Zhang|| 9:40|| 19:00|| 8+|| <br /> *finally ran the code on gpu, waiting for results to continue<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00 || 8+||<br /> * try to run rnng on GPU, but fail<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/9<br /> |Andy Zhang||10:00 ||19:00 ||8 || <br /> *bugs when doing validation, solved it but had to rerun the code<br /> *writing document of MemN2N<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00 || 8+||<br /> * try several wrong mem models<br /> * rebuild rnng on Dynet<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/10<br /> |Andy Zhang|| 9:30||18:30 || 8|| <br /> <br /> *writing document of MemN2N<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00|| 9||<br /> * read the code of Feng<br /> * try to run rnng on GPU, but fail<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/11<br /> |Andy Zhang|| 9:30||18:30 || 8|| <br /> *ran NTM on paraphrase data set<br /> <br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/14<br /> |Andy Zhang||9:30 ||18:30|| 8|| <br /> *prepared data set of fr-en for NMT; prepare training, val &amp; test set for paraphrase<br /> *ran NMT model on the above data<br /> *weekly meeting<br /> |-<br /> |Shiyue Zhang || 9:30 || 21:30|| 9+||<br /> * review last week work<br /> * meeting<br /> * try MKL, but it cannot use multi cpu cores<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/15<br /> |Andy Zhang||9:30 ||18:30|| 8|| <br /> *deal with paraphrase data set<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00 || 8+||<br /> * run rnng on MKL successfully, which can at least double the speed.<br /> * run the original rnng discriminative model<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/16<br /> |Andy Zhang||10:00 ||19:00|| 8|| <br /> *deal with paraphrase data set, wipe out repetitions and some noises<br /> *help Guli with NMT code<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00 || 8+||<br /> * finish the code of dynamic memory model and running<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/17<br /> |Andy Zhang||10:00 ||19:00||8 || <br /> *run NTM on data dealt yesterday<br /> *read through source code to find ways to modify it<br /> |-<br /> |Shiyue Zhang || 9:30 || 17:00|| 6 ||<br /> * try the memory structure mentioned by Teacher Wang<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/18<br /> |Andy Zhang||10:00 ||19:00||8 || <br /> *read through source code to find ways to modify it<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/21<br /> |Andy Zhang||9:30 ||19:00||8+ || <br /> *read source code and seem to figure out how to modify it<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/22<br /> |Andy Zhang||9:30 ||19:00||8+ || <br /> *deal with zh2en data set<br /> *run NTM on them<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/23<br /> |Andy Zhang||9:30 ||19:00||8+ || <br /> *discuss code with Mrs. Feng<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/24<br /> |Andy Zhang||10:00 ||19:00||8 || <br /> *met with crushes <br /> *figure out relationship between checkpoint model &amp; best model<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/25<br /> |Andy Zhang|||||||| <br /> <br /> |-<br /> |Shiyue Zhang || || || ||<br /> |}<br /> <br /> ===Monthly Summary===<br /> <br /> {| class=&quot;wikitable&quot;<br /> !People!! Summary<br /> |-<br /> |Yang Feng || <br /> |-<br /> |Jiyuan Zhang || <br /> |-<br /> |Andy Zhang || <br /> |-<br /> |Shiyue Zhang ||<br /> |}<br /> <br /> ===Time Off Table===<br /> <br /> {| class=&quot;wikitable&quot;<br /> ! Date !! Yang Feng !! Jiyuan Zhang <br /> |-<br /> |11/01 || 8h || <br /> |-<br /> |11/21 || 2h ||<br /> |}<br /> <br /> ==Past progress==<br /> [[nlp-progress 2016/10]]<br /> <br /> [[nlp-progress 2016/09]]<br /> <br /> [[nlp-progress 2016/08]]<br /> <br /> [[nlp-progress 2016/05/01 -- 08/16 | nlp-progress 2016/05-07]]<br /> <br /> [[nlp-progress 2016/04]]</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/Schedule 2016-11-21T11:54:57Z

<p>Zhangandy：/* Daily Report */</p> <hr /> <div>=NLP Schedule=<br /> <br /> ==Members==<br /> <br /> ===Current Members===<br /> <br /> * Yang Feng (冯洋)<br /> * Jiyuan Zhang （张记袁）<br /> * Aodong Li (李傲冬)<br /> * Andi Zhang (张安迪)<br /> * Aiting Liu (刘艾婷)<br /> * Shiyao Li （李诗瑶）<br /> * Shiyue Zhang (张诗悦)<br /> <br /> ===Former Members===<br /> * '''Chao Xing (邢超)''' : FreeNeb<br /> * '''Rong Liu (刘荣)''' : 优酷<br /> * '''Xiaoxi Wang (王晓曦)''' : 图灵机器人<br /> * '''Xi Ma (马习)''' : 清华大学研究生<br /> * '''Tianyi Luo (骆天一)''' ： phd candidate in University of California Santa Cruz<br /> * '''Qixin Wang (王琪鑫)''' : MA candidate in University of California<br /> * '''DongXu Zhang (张东旭)''': --<br /> * '''Yiqiao Pan (潘一桥)''' ： MA candidate in University of Sydney <br /> <br /> <br /> ==Work Progress==<br /> ===Daily Report===<br /> <br /> {|class=&quot;wikitable&quot;<br /> ! Date !! Person !! start!! leave !! hours ||status<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/1<br /> |Andy Zhang|| 10:00|| 19:00||8 || <br /> *read source code<br /> *weekly meeting <br /> |-<br /> |Shiyue Zhang || 9:15 || 19:30|| 8||<br /> * finish a simple way to add memory into rnng and write a report<br /> * meeting <br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/2<br /> |Andy Zhang|| 10:00|| 19:00||8 ||<br /> *tried to figure out the cost type of seq2seq<br /> *bimonthly report <br /> |-<br /> |Shiyue Zhang || 9:15 || 20:00|| 8+||<br /> * fix unexpected action<br /> * rerun original model<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/3<br /> |Andy Zhang|| 13:30|| 20:20|| 6|| <br /> *read source code <br /> |-<br /> |Shiyue Zhang || 9:35 || 19:00|| 8+||<br /> * rerun center memory model<br /> * run sample memory model<br /> * run wrong memory model<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/4<br /> |Andy Zhang|| 10:00|| 18:30|| 7+|| <br /> *try to run code on GPU but failed <br /> |-<br /> |Shiyue Zhang || |||| ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/7<br /> |Andy Zhang|| 10:00|| 19:20|| 8+|| <br /> *try to run code on GPU but failed <br /> |-<br /> |Shiyue Zhang || 9:15 || 20:30|| 9 ||<br /> * review last week work<br /> * write report <br /> * meeting<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/8<br /> |Andy Zhang|| 9:40|| 19:00|| 8+|| <br /> *finally ran the code on gpu, waiting for results to continue<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00 || 8+||<br /> * try to run rnng on GPU, but fail<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/9<br /> |Andy Zhang||10:00 ||19:00 ||8 || <br /> *bugs when doing validation, solved it but had to rerun the code<br /> *writing document of MemN2N<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00 || 8+||<br /> * try several wrong mem models<br /> * rebuild rnng on Dynet<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/10<br /> |Andy Zhang|| 9:30||18:30 || 8|| <br /> <br /> *writing document of MemN2N<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00|| 9||<br /> * read the code of Feng<br /> * try to run rnng on GPU, but fail<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/11<br /> |Andy Zhang|| 9:30||18:30 || 8|| <br /> *ran NTM on paraphrase data set<br /> <br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/14<br /> |Andy Zhang||9:30 ||18:30|| 8|| <br /> *prepared data set of fr-en for NMT; prepare training, val &amp; test set for paraphrase<br /> *ran NMT model on the above data<br /> *weekly meeting<br /> |-<br /> |Shiyue Zhang || 9:30 || 21:30|| 9+||<br /> * review last week work<br /> * meeting<br /> * try MKL, but it cannot use multi cpu cores<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/15<br /> |Andy Zhang||9:30 ||18:30|| 8|| <br /> *deal with paraphrase data set<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00 || 8+||<br /> * run rnng on MKL successfully, which can at least double the speed.<br /> * run the original rnng discriminative model<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/16<br /> |Andy Zhang||10:00 ||19:00|| 8|| <br /> *deal with paraphrase data set, wipe out repetitions and some noises<br /> *help Guli with NMT code<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00 || 8+||<br /> * finish the code of dynamic memory model and running<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/17<br /> |Andy Zhang||10:00 ||19:00||8 || <br /> *run NTM on data dealt yesterday<br /> *read through source code to find ways to modify it<br /> |-<br /> |Shiyue Zhang || 9:30 || 17:00|| 6 ||<br /> * try the memory structure mentioned by Teacher Wang<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/18<br /> |Andy Zhang||10:00 ||19:00||8 || <br /> *read through source code to find ways to modify it<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/18<br /> |Andy Zhang||9:30 ||19:00||8+ || <br /> *read source code and seem to figure out how to modify it<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |}<br /> <br /> ===Monthly Summary===<br /> <br /> {| class=&quot;wikitable&quot;<br /> !People!! Summary<br /> |-<br /> |Yang Feng || <br /> |-<br /> |Jiyuan Zhang || <br /> |-<br /> |Andy Zhang || <br /> |-<br /> |Shiyue Zhang ||<br /> |}<br /> <br /> ===Time Off Table===<br /> <br /> {| class=&quot;wikitable&quot;<br /> ! Date !! Yang Feng !! Jiyuan Zhang <br /> |-<br /> |11/01 || 8h || <br /> |}<br /> <br /> ==Past progress==<br /> [[nlp-progress 2016/10]]<br /> <br /> [[nlp-progress 2016/09]]<br /> <br /> [[nlp-progress 2016/08]]<br /> <br /> [[nlp-progress 2016/05/01 -- 08/16 | nlp-progress 2016/05-07]]<br /> <br /> [[nlp-progress 2016/04]]</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/NLP_Status_Report_2016-11-21 2016-11-20T12:46:34Z

<p>Zhangandy：</p> <hr /> <div>{| class=&quot;wikitable&quot;<br /> !Date !! People !! Last Week !! This Week<br /> |-<br /> | rowspan=&quot;5&quot;|2016/11/21<br /> |Yang Feng ||<br /> <br /> ||<br /> <br /> |-<br /> |Jiyuan Zhang ||<br /> <br /> || <br /> <br /> |-<br /> |Andi Zhang ||<br /> *prepare new data set for paraphrase, wiped out repetition &amp; most of the noises<br /> *run NMT on fr-en data set and new paraphrase set<br /> *read through source code to find ways to modify it<br /> *helped Guli with running NMT on our server<br /> ||<br /> *decide to drop theano or not<br /> *start to work on codes<br /> |-<br /> |Shiyue Zhang || <br /> <br /> ||<br /> <br /> |-<br /> |Guli ||<br /> <br /> ||<br /> <br /> |}</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/NLP_Status_Report_2016-11-21 2016-11-20T12:00:36Z

<p>Zhangandy：以“{| class=&quot;wikitable&quot; !Date !! People !! Last Week !! This Week |- | rowspan=&quot;5&quot;|2016/11/21 |Yang Feng || || |- |Jiyuan Zhang || || |- |Andi Zhang || || |- |S...”为内容创建页面</p> <hr /> <div>{| class=&quot;wikitable&quot;<br /> !Date !! People !! Last Week !! This Week<br /> |-<br /> | rowspan=&quot;5&quot;|2016/11/21<br /> |Yang Feng ||<br /> <br /> ||<br /> <br /> |-<br /> |Jiyuan Zhang ||<br /> <br /> || <br /> <br /> |-<br /> |Andi Zhang ||<br /> <br /> <br /> ||<br /> <br /> |-<br /> |Shiyue Zhang || <br /> <br /> ||<br /> <br /> |-<br /> |Guli ||<br /> <br /> ||<br /> <br /> |}</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/2016-11-21 2016-11-20T11:59:08Z

<p>Zhangandy：以“NLP Status Report 2016-11-21 ASR Status Report 2016-11-21 FIN Status Report 2016-11-21 FreeNeb Status Report 2016-11-21”为内容创建页面</p> <hr /> <div>[[NLP Status Report 2016-11-21]]<br /> <br /> [[ASR Status Report 2016-11-21]]<br /> <br /> [[FIN Status Report 2016-11-21]]<br /> <br /> [[FreeNeb Status Report 2016-11-21]]</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/Status_report 2016-11-20T11:58:41Z

<p>Zhangandy：</p> <hr /> <div>[[2016-11-21]]<br /> <br /> [[2016-11-14]]<br /> <br /> [[2016-11-07]]<br /> <br /> [[2016-10-31]]<br /> <br /> [[2016-10-24]]<br /> <br /> [[2016-10-17]]<br /> <br /> [[2016-10-10]] <br /> <br /> [[2016-09-26]]<br /> <br /> [[2016-09-19]]<br /> <br /> [[2016-09-12]]<br /> <br /> [[2016-09-05]]<br /> <br /> [[2016-08-29]]<br /> <br /> [[2016-08-22]]<br /> <br /> [[2016-08-15]]<br /> <br /> [[2016-08-08]]<br /> <br /> [[2016-08-01]]<br /> <br /> [[2016-07-25]]<br /> <br /> [[2016-07-18]]<br /> <br /> [[2016-07-11]]<br /> <br /> [[2016-07-04]]<br /> <br /> [[2016-06-27]]<br /> <br /> [[2016-06-19]]<br /> <br /> [[2016-06-13]]<br /> <br /> [[2016-06-06]]<br /> <br /> [[2016-05-30]]<br /> <br /> [[2016-05-23]]<br /> <br /> [[2016-05-16]]<br /> <br /> [[2016-05-09]]<br /> <br /> [[2016-05-02]]<br /> <br /> [[2016-04-25]]<br /> <br /> [[2016-04-18]]<br /> <br /> [[2016-04-11]]<br /> <br /> [[2016-04-04]]<br /> <br /> [[2016-03-28]]<br /> <br /> [[2016-03-21]]<br /> <br /> [[2016-03-14]]<br /> <br /> [[2016-03-07]]<br /> <br /> [[2016-02-29]]<br /> <br /> [[2016-02-22]]<br /> <br /> [[2016-02-01]]<br /> <br /> [[2016-01-18]]<br /> <br /> [[2016-01-25]]<br /> <br /> [[2016-01-11]]<br /> <br /> [[2016-01-04]]</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/Schedule 2016-11-20T11:58:17Z

<p>Zhangandy：/* Daily Report */</p> <hr /> <div>=NLP Schedule=<br /> <br /> ==Members==<br /> <br /> ===Current Members===<br /> <br /> * Yang Feng (冯洋)<br /> * Jiyuan Zhang （张记袁）<br /> * Aodong Li (李傲冬)<br /> * Andi Zhang (张安迪)<br /> * Aiting Liu (刘艾婷)<br /> * Shiyao Li （李诗瑶）<br /> * Shiyue Zhang (张诗悦)<br /> <br /> ===Former Members===<br /> * '''Chao Xing (邢超)''' : FreeNeb<br /> * '''Rong Liu (刘荣)''' : 优酷<br /> * '''Xiaoxi Wang (王晓曦)''' : 图灵机器人<br /> * '''Xi Ma (马习)''' : 清华大学研究生<br /> * '''Tianyi Luo (骆天一)''' ： phd candidate in University of California Santa Cruz<br /> * '''Qixin Wang (王琪鑫)''' : MA candidate in University of California<br /> * '''DongXu Zhang (张东旭)''': --<br /> * '''Yiqiao Pan (潘一桥)''' ： MA candidate in University of Sydney <br /> <br /> <br /> ==Work Progress==<br /> ===Daily Report===<br /> <br /> {|class=&quot;wikitable&quot;<br /> ! Date !! Person !! start!! leave !! hours ||status<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/1<br /> |Andy Zhang|| 10:00|| 19:00||8 || <br /> *read source code<br /> *weekly meeting <br /> |-<br /> |Shiyue Zhang || 9:15 || 19:30|| 8||<br /> * finish a simple way to add memory into rnng and write a report<br /> * meeting <br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/2<br /> |Andy Zhang|| 10:00|| 19:00||8 ||<br /> *tried to figure out the cost type of seq2seq<br /> *bimonthly report <br /> |-<br /> |Shiyue Zhang || 9:15 || 20:00|| 8+||<br /> * fix unexpected action<br /> * rerun original model<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/3<br /> |Andy Zhang|| 13:30|| 20:20|| 6|| <br /> *read source code <br /> |-<br /> |Shiyue Zhang || 9:35 || 19:00|| 8+||<br /> * rerun center memory model<br /> * run sample memory model<br /> * run wrong memory model<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/4<br /> |Andy Zhang|| 10:00|| 18:30|| 7+|| <br /> *try to run code on GPU but failed <br /> |-<br /> |Shiyue Zhang || |||| ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/7<br /> |Andy Zhang|| 10:00|| 19:20|| 8+|| <br /> *try to run code on GPU but failed <br /> |-<br /> |Shiyue Zhang || 9:15 || 20:30|| 9 ||<br /> * review last week work<br /> * write report <br /> * meeting<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/8<br /> |Andy Zhang|| 9:40|| 19:00|| 8+|| <br /> *finally ran the code on gpu, waiting for results to continue<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00 || 8+||<br /> * try to run rnng on GPU, but fail<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/9<br /> |Andy Zhang||10:00 ||19:00 ||8 || <br /> *bugs when doing validation, solved it but had to rerun the code<br /> *writing document of MemN2N<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00 || 8+||<br /> * try several wrong mem models<br /> * rebuild rnng on Dynet<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/10<br /> |Andy Zhang|| 9:30||18:30 || 8|| <br /> <br /> *writing document of MemN2N<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00|| 9||<br /> * read the code of Feng<br /> * try to run rnng on GPU, but fail<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/11<br /> |Andy Zhang|| 9:30||18:30 || 8|| <br /> *ran NTM on paraphrase data set<br /> <br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/14<br /> |Andy Zhang||9:30 ||18:30|| 8|| <br /> *prepared data set of fr-en for NMT; prepare training, val &amp; test set for paraphrase<br /> *ran NMT model on the above data<br /> *weekly meeting<br /> |-<br /> |Shiyue Zhang || 9:30 || 21:30|| 9+||<br /> * review last week work<br /> * meeting<br /> * try MKL, but it cannot use multi cpu cores<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/15<br /> |Andy Zhang||9:30 ||18:30|| 8|| <br /> *deal with paraphrase data set<br /> |-<br /> |Shiyue Zhang || || || ||<br /> <br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/16<br /> |Andy Zhang||10:00 ||19:00|| 8|| <br /> *deal with paraphrase data set, wipe out repetitions and some noises<br /> *help Guli with NMT code<br /> |-<br /> |Shiyue Zhang || |||| ||<br /> <br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/17<br /> |Andy Zhang||10:00 ||19:00||8 || <br /> *run NTM on data dealt yesterday<br /> *read through source code to find ways to modify it<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/18<br /> |Andy Zhang||10:00 ||19:00||8 || <br /> *read through source code to find ways to modify it<br /> |-<br /> |Shiyue Zhang || || || ||<br /> |}<br /> <br /> ===Monthly Summary===<br /> <br /> {| class=&quot;wikitable&quot;<br /> !People!! Summary<br /> |-<br /> |Yang Feng || <br /> |-<br /> |Jiyuan Zhang || <br /> |-<br /> |Andy Zhang || <br /> |-<br /> |Shiyue Zhang ||<br /> |}<br /> <br /> ===Time Off Table===<br /> <br /> {| class=&quot;wikitable&quot;<br /> ! Date !! Yang Feng !! Jiyuan Zhang <br /> |-<br /> |11/01 || 8h || <br /> |}<br /> <br /> ==Past progress==<br /> [[nlp-progress 2016/10]]<br /> <br /> [[nlp-progress 2016/09]]<br /> <br /> [[nlp-progress 2016/08]]<br /> <br /> [[nlp-progress 2016/05/01 -- 08/16 | nlp-progress 2016/05-07]]<br /> <br /> [[nlp-progress 2016/04]]</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/Schedule 2016-11-17T02:35:24Z

<p>Zhangandy：/* Daily Report */</p> <hr /> <div>=NLP Schedule=<br /> <br /> ==Members==<br /> <br /> ===Current Members===<br /> <br /> * Yang Feng (冯洋)<br /> * Jiyuan Zhang （张记袁）<br /> * Aodong Li (李傲冬)<br /> * Andi Zhang (张安迪)<br /> * Aiting Liu (刘艾婷)<br /> * Shiyao Li （李诗瑶）<br /> * Shiyue Zhang (张诗悦)<br /> <br /> ===Former Members===<br /> * '''Chao Xing (邢超)''' : FreeNeb<br /> * '''Rong Liu (刘荣)''' : 优酷<br /> * '''Xiaoxi Wang (王晓曦)''' : 图灵机器人<br /> * '''Xi Ma (马习)''' : 清华大学研究生<br /> * '''Tianyi Luo (骆天一)''' ： phd candidate in University of California Santa Cruz<br /> * '''Qixin Wang (王琪鑫)''' : MA candidate in University of California<br /> * '''DongXu Zhang (张东旭)''': --<br /> * '''Yiqiao Pan (潘一桥)''' ： MA candidate in University of Sydney <br /> <br /> <br /> ==Work Progress==<br /> ===Daily Report===<br /> <br /> {|class=&quot;wikitable&quot;<br /> ! Date !! Person !! start!! leave !! hours ||status<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/1<br /> |Andy Zhang|| 10:00|| 19:00||8 || <br /> *read source code<br /> *weekly meeting <br /> |-<br /> |Shiyue Zhang || 9:15 || 19:30|| 8||<br /> * finish a simple way to add memory into rnng and write a report<br /> * meeting <br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/2<br /> |Andy Zhang|| 10:00|| 19:00||8 ||<br /> *tried to figure out the cost type of seq2seq<br /> *bimonthly report <br /> |-<br /> |Shiyue Zhang || 9:15 || 20:00|| 8+||<br /> * fix unexpected action<br /> * rerun original model<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/3<br /> |Andy Zhang|| 13:30|| 20:20|| 6|| <br /> *read source code <br /> |-<br /> |Shiyue Zhang || 9:35 || 19:00|| 8+||<br /> * rerun center memory model<br /> * run sample memory model<br /> * run wrong memory model<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/4<br /> |Andy Zhang|| 10:00|| 18:30|| 7+|| <br /> *try to run code on GPU but failed <br /> |-<br /> |Shiyue Zhang || |||| ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/7<br /> |Andy Zhang|| 10:00|| 19:20|| 8+|| <br /> *try to run code on GPU but failed <br /> |-<br /> |Shiyue Zhang || 9:15 || 20:30|| 9 ||<br /> * review last week work<br /> * write report <br /> * meeting<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/8<br /> |Andy Zhang|| 9:40|| 19:00|| 8+|| <br /> *finally ran the code on gpu, waiting for results to continue<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00 || 8+||<br /> * try to run rnng on GPU, but fail<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/9<br /> |Andy Zhang||10:00 ||19:00 ||8 || <br /> *bugs when doing validation, solved it but had to rerun the code<br /> *writing document of MemN2N<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00 || 8+||<br /> * try several wrong mem models<br /> * rebuild rnng on Dynet<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/10<br /> |Andy Zhang|| 9:30||18:30 || 8|| <br /> <br /> *writing document of MemN2N<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00|| 9||<br /> * read the code of Feng<br /> * try to run rnng on GPU, but fail<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/11<br /> |Andy Zhang|| 9:30||18:30 || 8|| <br /> *ran NTM on paraphrase data set<br /> <br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/14<br /> |Andy Zhang||9:30 ||18:30|| 8|| <br /> *prepared data set of fr-en for NMT; prepare training, val &amp; test set for paraphrase<br /> *ran NMT model on the above data<br /> *weekly meeting<br /> |-<br /> |Shiyue Zhang || 9:30 || 21:30|| 9+||<br /> * review last week work<br /> * meeting<br /> * try MKL, but it cannot use multi cpu cores<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/15<br /> |Andy Zhang||9:30 ||18:30|| 8|| <br /> *deal with paraphrase data set<br /> |-<br /> |Shiyue Zhang || || || ||<br /> <br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/16<br /> |Andy Zhang||10:00 ||19:00|| 8|| <br /> *deal with paraphrase data set, wipe out repetitions and some noises<br /> *help Guli with NMT code<br /> |-<br /> |Shiyue Zhang || |||| ||<br /> <br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/17<br /> |Andy Zhang|| |||| || <br /> <br /> |-<br /> |Shiyue Zhang || || || ||<br /> <br /> |}<br /> <br /> ===Monthly Summary===<br /> <br /> {| class=&quot;wikitable&quot;<br /> !People!! Summary<br /> |-<br /> |Yang Feng || <br /> |-<br /> |Jiyuan Zhang || <br /> |-<br /> |Andy Zhang || <br /> |-<br /> |Shiyue Zhang ||<br /> |}<br /> <br /> ===Time Off Table===<br /> <br /> {| class=&quot;wikitable&quot;<br /> ! Date !! Yang Feng !! Jiyuan Zhang <br /> |-<br /> |11/01 || 8h || <br /> |}<br /> <br /> ==Past progress==<br /> [[nlp-progress 2016/10]]<br /> <br /> [[nlp-progress 2016/09]]<br /> <br /> [[nlp-progress 2016/08]]<br /> <br /> [[nlp-progress 2016/05/01 -- 08/16 | nlp-progress 2016/05-07]]<br /> <br /> [[nlp-progress 2016/04]]</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/Schedule 2016-11-15T02:20:28Z

<p>Zhangandy：/* Daily Report */</p> <hr /> <div>=NLP Schedule=<br /> <br /> ==Members==<br /> <br /> ===Current Members===<br /> <br /> * Yang Feng (冯洋)<br /> * Jiyuan Zhang （张记袁）<br /> * Aodong Li (李傲冬)<br /> * Andi Zhang (张安迪)<br /> * Aiting Liu (刘艾婷)<br /> * Shiyao Li （李诗瑶）<br /> * Shiyue Zhang (张诗悦)<br /> <br /> ===Former Members===<br /> * '''Chao Xing (邢超)''' : FreeNeb<br /> * '''Rong Liu (刘荣)''' : 优酷<br /> * '''Xiaoxi Wang (王晓曦)''' : 图灵机器人<br /> * '''Xi Ma (马习)''' : 清华大学研究生<br /> * '''Tianyi Luo (骆天一)''' ： phd candidate in University of California Santa Cruz<br /> * '''Qixin Wang (王琪鑫)''' : MA candidate in University of California<br /> * '''DongXu Zhang (张东旭)''': --<br /> * '''Yiqiao Pan (潘一桥)''' ： MA candidate in University of Sydney <br /> <br /> <br /> ==Work Progress==<br /> ===Daily Report===<br /> <br /> {|class=&quot;wikitable&quot;<br /> ! Date !! Person !! start!! leave !! hours ||status<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/1<br /> |Andy Zhang|| 10:00|| 19:00||8 || <br /> *read source code<br /> *weekly meeting <br /> |-<br /> |Shiyue Zhang || 9:15 || 19:30|| 8||<br /> * finish a simple way to add memory into rnng and write a report<br /> * meeting <br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/2<br /> |Andy Zhang|| 10:00|| 19:00||8 ||<br /> *tried to figure out the cost type of seq2seq<br /> *bimonthly report <br /> |-<br /> |Shiyue Zhang || 9:15 || 20:00|| 8+||<br /> * fix unexpected action<br /> * rerun original model<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/3<br /> |Andy Zhang|| 13:30|| 20:20|| 6|| <br /> *read source code <br /> |-<br /> |Shiyue Zhang || 9:35 || 19:00|| 8+||<br /> * rerun center memory model<br /> * run sample memory model<br /> * run wrong memory model<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/4<br /> |Andy Zhang|| 10:00|| 18:30|| 7+|| <br /> *try to run code on GPU but failed <br /> |-<br /> |Shiyue Zhang || |||| ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/7<br /> |Andy Zhang|| 10:00|| 19:20|| 8+|| <br /> *try to run code on GPU but failed <br /> |-<br /> |Shiyue Zhang || 9:15 || 20:30|| 9 ||<br /> * review last week work<br /> * write report <br /> * meeting<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/8<br /> |Andy Zhang|| 9:40|| 19:00|| 8+|| <br /> *finally ran the code on gpu, waiting for results to continue<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00 || 8+||<br /> * try to run rnng on GPU, but fail<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/9<br /> |Andy Zhang||10:00 ||19:00 ||8 || <br /> *bugs when doing validation, solved it but had to rerun the code<br /> *writing document of MemN2N<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00 || 8+||<br /> * try several wrong mem models<br /> * rebuild rnng on Dynet<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/10<br /> |Andy Zhang|| 9:30||18:30 || 8|| <br /> <br /> *writing document of MemN2N<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00|| 9||<br /> * read the code of Feng<br /> * try to run rnng on GPU, but fail<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/11<br /> |Andy Zhang|| 9:30||18:30 || 8|| <br /> *ran NTM on paraphrase data set<br /> <br /> |-<br /> |Shiyue Zhang || || || ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/14<br /> |Andy Zhang||9:30 ||18:30|| 8|| <br /> *prepared data set of fr-en for NMT; prepare training, val &amp; test set for paraphrase<br /> *ran NMT model on the above data<br /> *weekly meeting<br /> |-<br /> |Shiyue Zhang || 9:30 || 21:30|| 9+||<br /> * review last week work<br /> * meeting<br /> * try MKL, but it cannot use multi cpu cores<br /> |}<br /> <br /> ===Monthly Summary===<br /> <br /> {| class=&quot;wikitable&quot;<br /> !People!! Summary<br /> |-<br /> |Yang Feng || <br /> |-<br /> |Jiyuan Zhang || <br /> |-<br /> |Andy Zhang || <br /> |-<br /> |Shiyue Zhang ||<br /> |}<br /> <br /> ===Time Off Table===<br /> <br /> {| class=&quot;wikitable&quot;<br /> ! Date !! Yang Feng !! Jiyuan Zhang <br /> |-<br /> |11/01 || 8h || <br /> |}<br /> <br /> ==Past progress==<br /> [[nlp-progress 2016/10]]<br /> <br /> [[nlp-progress 2016/09]]<br /> <br /> [[nlp-progress 2016/08]]<br /> <br /> [[nlp-progress 2016/05/01 -- 08/16 | nlp-progress 2016/05-07]]<br /> <br /> [[nlp-progress 2016/04]]</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/Asr-language-processing-research-s2s-generation 2016-11-14T07:23:37Z

<p>Zhangandy：/* Time Table */</p> <hr /> <div>==Main Idea==<br /> <br /> ==People==<br /> Yang Feng, Andi Zhang<br /> <br /> ==Time Table==<br /> {| class=&quot;wikitable&quot;<br /> !Week !! Work Plan !! Work Done <br /> |-<br /> | 2016/11/07-2016/11/13 ||<br /> *successfully run the code of &quot;Neural machine translation by jointly learning to align and translate&quot; on gpu<br /> *start working on model_step_1: linear trans-&gt;cosine-&gt;linear trans-&gt;softmax; start coding if time permits. <br /> ||90%<br /> |-<br /> | 2016/11/14-2016/11/20||<br /> *coding on model_step_1<br /> *run &amp; test the code<br /> *run NTM on fr-en data set<br /> *run NTM on paraphrase data set<br /> ||<br /> |-<br /> | 2016/11/21-2016/11/27 || <br /> *continue working on model_step_1<br /> *start working on model_step_2: lstm-&gt;mn-&gt;softmax<br /> ||<br /> |-<br /> | 2016/11/28-2016/12/04 || <br /> *coding &amp; debug &amp; run model_step_2<br /> *start working on model_step_3: joint training<br /> ||<br /> |-<br /> | 2016/12/05-2016/12/11 ||<br /> *coding &amp; debug &amp; run model_step_3<br /> <br /> ||<br /> |-<br /> | 2016/12/12-2016/12/18 || <br /> *find ways to speed up the model if it is slow<br /> ||<br /> |-<br /> | 2016/12/19-2016/12/25 || <br /> *select memory to optimize result<br /> ||<br /> |-<br /> | 2016/12/26-2016/12/31 || <br /> *run final result<br /> *check any possible faults<br /> ||<br /> |-<br /> <br /> |}<br /> <br /> ==Progress==</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/Schedule 2016-11-14T07:21:21Z

<p>Zhangandy：/* Daily Report */</p> <hr /> <div>=NLP Schedule=<br /> <br /> ==Members==<br /> <br /> ===Current Members===<br /> <br /> * Yang Feng (冯洋)<br /> * Jiyuan Zhang （张记袁）<br /> * Aodong Li (李傲冬)<br /> * Andi Zhang (张安迪)<br /> * Aiting Liu (刘艾婷)<br /> * Shiyao Li （李诗瑶）<br /> * Shiyue Zhang (张诗悦)<br /> <br /> ===Former Members===<br /> * '''Chao Xing (邢超)''' : FreeNeb<br /> * '''Rong Liu (刘荣)''' : 优酷<br /> * '''Xiaoxi Wang (王晓曦)''' : 图灵机器人<br /> * '''Xi Ma (马习)''' : 清华大学研究生<br /> * '''Tianyi Luo (骆天一)''' ： phd candidate in University of California Santa Cruz<br /> * '''Qixin Wang (王琪鑫)''' : MA candidate in University of California<br /> * '''DongXu Zhang (张东旭)''': --<br /> * '''Yiqiao Pan (潘一桥)''' ： MA candidate in University of Sydney <br /> <br /> <br /> ==Work Progress==<br /> ===Daily Report===<br /> <br /> {|class=&quot;wikitable&quot;<br /> ! Date !! Person !! start!! leave !! hours ||status<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/1<br /> |Andy Zhang|| 10:00|| 19:00||8 || <br /> *read source code<br /> *weekly meeting <br /> |-<br /> |Shiyue Zhang || 9:15 || 19:30|| 8||<br /> * finish a simple way to add memory into rnng and write a report<br /> * meeting <br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/2<br /> |Andy Zhang|| 10:00|| 19:00||8 ||<br /> *tried to figure out the cost type of seq2seq<br /> *bimonthly report <br /> |-<br /> |Shiyue Zhang || 9:15 || 20:00|| 8+||<br /> * fix unexpected action<br /> * rerun original model<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/3<br /> |Andy Zhang|| 13:30|| 20:20|| 6|| <br /> *read source code <br /> |-<br /> |Shiyue Zhang || 9:35 || 19:00|| 8+||<br /> * rerun center memory model<br /> * run sample memory model<br /> * run wrong memory model<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/4<br /> |Andy Zhang|| 10:00|| 18:30|| 7+|| <br /> *try to run code on GPU but failed <br /> |-<br /> |Shiyue Zhang || |||| ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/7<br /> |Andy Zhang|| 10:00|| 19:20|| 8+|| <br /> *try to run code on GPU but failed <br /> |-<br /> |Shiyue Zhang || 9:15 || 20:30|| 9 ||<br /> * review last week work<br /> * write report <br /> * meeting<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/8<br /> |Andy Zhang|| 9:40|| 19:00|| 8+|| <br /> *finally ran the code on gpu, waiting for results to continue<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00 || 8+||<br /> * try to run rnng on GPU, but fail<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/9<br /> |Andy Zhang||10:00 ||19:00 ||8 || <br /> *bugs when doing validation, solved it but had to rerun the code<br /> *writing document of MemN2N<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00 || 8+||<br /> * try several wrong mem models<br /> * rebuild rnng on Dynet<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/10<br /> |Andy Zhang|| 9:30||18:30 || 8|| <br /> <br /> *writing document of MemN2N<br /> |-<br /> |Shiyue Zhang || 9:30 || 20:00|| 9||<br /> * read the code of Feng<br /> * try to run rnng on GPU, but fail<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/11<br /> |Andy Zhang|| 9:30||18:30 || 8|| <br /> *ran NTM on paraphrase data set<br /> <br /> |-<br /> |Shiyue Zhang || || || ||<br /> <br /> |}<br /> <br /> ===Monthly Summary===<br /> <br /> {| class=&quot;wikitable&quot;<br /> !People!! Summary<br /> |-<br /> |Yang Feng || <br /> |-<br /> |Jiyuan Zhang || <br /> |-<br /> |Andy Zhang || <br /> |-<br /> |Shiyue Zhang ||<br /> |}<br /> <br /> ===Time Off Table===<br /> <br /> {| class=&quot;wikitable&quot;<br /> ! Date !! Yang Feng !! Jiyuan Zhang <br /> |-<br /> |11/01 || 8h || <br /> |}<br /> <br /> ==Past progress==<br /> [[nlp-progress 2016/10]]<br /> <br /> [[nlp-progress 2016/09]]<br /> <br /> [[nlp-progress 2016/08]]<br /> <br /> [[nlp-progress 2016/05/01 -- 08/16 | nlp-progress 2016/05-07]]<br /> <br /> [[nlp-progress 2016/04]]</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/NLP_Status_Report_2016-11-14 2016-11-14T04:43:30Z

<p>Zhangandy：</p> <hr /> <div>{| class=&quot;wikitable&quot;<br /> !Date !! People !! Last Week !! This Week<br /> |-<br /> | rowspan=&quot;4&quot;|2016/10/31<br /> |Yang Feng ||<br /> * added new features to rnng+mn, including automatically detecting wrong sentences, swapping memories more frequently and filtering memory units to speed up.<br /> * ran experiments for rnng+mn [[http://cslt.riit.tsinghua.edu.cn/mediawiki/images/f/f8/Progress_of_RNNG_with_memory_network.pdf report]]<br /> * read the code of sequence-to-sequence with tensorflow<br /> * recruited interns<br /> * Huilan work summary<br /> ||<br /> * optimize rnng+MN;<br /> * discuss the code with Jiyuan;<br /> * work with Andy at NMT;<br /> * Intern interviews<br /> * Huilan work.<br /> |-<br /> |Jiyuan Zhang ||<br /> <br /> || <br /> <br /> |-<br /> |Andi Zhang ||<br /> *ran NMT (cs-en) on gpu, but bleu is low, could be resulting from a small corpus<br /> *ran NMT on paraphrase data set<br /> *wrote MemN2N ducument <br /> <br /> ||<br /> *run NMT (fr-en) to get a bleu as that in the paper<br /> *run paraphrase for validation<br /> |-<br /> |Shiyue Zhang || <br /> <br /> ||<br /> <br /> |-<br /> |}</div>

Zhangandy http://index.cslt.org/mediawiki/index.php/Schedule 2016-11-11T01:13:44Z

<p>Zhangandy：/* Daily Report */</p> <hr /> <div>=NLP Schedule=<br /> <br /> ==Members==<br /> <br /> ===Current Members===<br /> <br /> * Yang Feng (冯洋)<br /> * Jiyuan Zhang （张记袁）<br /> * Aodong Li (李傲冬)<br /> * Andi Zhang (张安迪)<br /> * Aiting Liu (刘艾婷)<br /> * Shiyao Li （李诗瑶）<br /> * Shiyue Zhang (张诗悦)<br /> <br /> ===Former Members===<br /> * '''Chao Xing (邢超)''' : FreeNeb<br /> * '''Rong Liu (刘荣)''' : 优酷<br /> * '''Xiaoxi Wang (王晓曦)''' : 图灵机器人<br /> * '''Xi Ma (马习)''' : 清华大学研究生<br /> * '''Tianyi Luo (骆天一)''' ： phd candidate in University of California Santa Cruz<br /> * '''Qixin Wang (王琪鑫)''' : MA candidate in University of California<br /> * '''DongXu Zhang (张东旭)''': --<br /> * '''Yiqiao Pan (潘一桥)''' ： MA candidate in University of Sydney <br /> <br /> <br /> ==Work Progress==<br /> ===Daily Report===<br /> <br /> {|class=&quot;wikitable&quot;<br /> ! Date !! Person !! start!! leave !! hours ||status<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/1<br /> |Andy Zhang|| 10:00|| 19:00||8 || <br /> *read source code<br /> *weekly meeting <br /> |-<br /> |Shiyue Zhang || 9:15 || 19:30|| 8||<br /> * finish a simple way to add memory into rnng and write a report<br /> * meeting <br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/2<br /> |Andy Zhang|| 10:00|| 19:00||8 ||<br /> *tried to figure out the cost type of seq2seq<br /> *bimonthly report <br /> |-<br /> |Shiyue Zhang || 9:15 || 20:00|| 8+||<br /> * fix unexpected action<br /> * rerun original model<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/3<br /> |Andy Zhang|| 13:30|| 20:20|| 6|| <br /> *read source code <br /> |-<br /> |Shiyue Zhang || 9:35 || 19:00|| 8+||<br /> * rerun center memory model<br /> * run sample memory model<br /> * run wrong memory model<br /> |-<br /> |Shiyue Zhang || |||| ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/4<br /> |Andy Zhang|| 10:00|| 18:30|| 7+|| <br /> *try to run code on GPU but failed <br /> |-<br /> |Shiyue Zhang || |||| ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/7<br /> |Andy Zhang|| 10:00|| 19:20|| 8+|| <br /> *try to run code on GPU but failed <br /> |-<br /> |Shiyue Zhang || |||| ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/8<br /> |Andy Zhang|| 9:40|| 19:00|| 8+|| <br /> *finally ran the code on gpu, waiting for results to continue<br /> |-<br /> |Shiyue Zhang || |||| ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/9<br /> |Andy Zhang||10:00 ||19:00 ||8 || <br /> *bugs when doing validation, solved it but had to rerun the code<br /> *writing document of MemN2N<br /> |-<br /> |Shiyue Zhang || |||| ||<br /> |-<br /> | rowspan=&quot;2&quot;|2016/11/10<br /> |Andy Zhang|| 9:30||18:30 || 8|| <br /> <br /> *writing document of MemN2N<br /> |-<br /> |Shiyue Zhang || |||| ||<br /> |}<br /> <br /> ===Monthly Summary===<br /> <br /> {| class=&quot;wikitable&quot;<br /> !People!! Summary<br /> |-<br /> |Yang Feng || <br /> |-<br /> |Jiyuan Zhang || <br /> |-<br /> |Andy Zhang || <br /> |-<br /> |Shiyue Zhang ||<br /> |}<br /> <br /> ===Time Off Table===<br /> <br /> {| class=&quot;wikitable&quot;<br /> ! Date !! Yang Feng !! Jiyuan Zhang <br /> |-<br /> |11/01 || 8h || <br /> |}<br /> <br /> ==Past progress==<br /> [[nlp-progress 2016/10]]<br /> <br /> [[nlp-progress 2016/09]]<br /> <br /> [[nlp-progress 2016/08]]<br /> <br /> [[nlp-progress 2016/05/01 -- 08/16 | nlp-progress 2016/05-07]]<br /> <br /> [[nlp-progress 2016/04]]</div>

Zhangandy