“Asr-language-processing-research-s2s-generation”版本间的差异

来自cslt Wiki
跳转至： 导航、 搜索
（→‎Time Table）
（→‎Time Table）
 
第11行： 第11行：
 
*successfully run the code of "Neural machine translation by jointly learning to align and translate" on gpu
 
*successfully run the code of "Neural machine translation by jointly learning to align and translate" on gpu
 
*start working on model_step_1: linear trans->cosine->linear trans->softmax; start coding if time permits.  
 
*start working on model_step_1: linear trans->cosine->linear trans->softmax; start coding if time permits.  
−
||
+
||90%
 
|-
 
|-
 
| 2016/11/14-2016/11/20||
 
| 2016/11/14-2016/11/20||
 
*coding on model_step_1
 
*coding on model_step_1
 
*run & test the code
 
*run & test the code
 +
*run NTM on fr-en data set
 +
*run NTM on paraphrase data set
 
  ||
 
  ||
 
|-
 
|-

2016年11月14日 (一) 07:23的最后版本

Main Idea

People

Yang Feng, Andi Zhang

Time Table

Week Work Plan Work Done
2016/11/07-2016/11/13
  • successfully run the code of "Neural machine translation by jointly learning to align and translate" on gpu
  • start working on model_step_1: linear trans->cosine->linear trans->softmax; start coding if time permits.
90%
2016/11/14-2016/11/20
  • coding on model_step_1
  • run & test the code
  • run NTM on fr-en data set
  • run NTM on paraphrase data set
2016/11/21-2016/11/27
  • continue working on model_step_1
  • start working on model_step_2: lstm->mn->softmax
2016/11/28-2016/12/04
  • coding & debug & run model_step_2
  • start working on model_step_3: joint training
2016/12/05-2016/12/11
  • coding & debug & run model_step_3
2016/12/12-2016/12/18
  • find ways to speed up the model if it is slow
2016/12/19-2016/12/25
  • select memory to optimize result
2016/12/26-2016/12/31
  • run final result
  • check any possible faults

Progress

取自“http://index.cslt.org/mediawiki/index.php?title=Asr-language-processing-research-s2s-generation&oldid=23847”