“Huilian-cslt-week-2016-3-03”版本间的差异
来自cslt Wiki
(以“==Environment== ==RNN Poem Processing== * contact specialists --Qixin done * speed up training(minibath parallel) done * update a new version ==RNN Writing Pro...”为内容创建页面) |
|||
第1行: | 第1行: | ||
==Environment== | ==Environment== | ||
− | ==RNN Poem Processing== | + | ==RNN Poem Processing== |
− | + | * update a new version -Qixin | |
− | + | ||
− | * update a new version | + | |
==RNN Writing Processing== | ==RNN Writing Processing== | ||
第13行: | 第11行: | ||
==RNN QA== | ==RNN QA== | ||
* attention based QA (Solution for template expansion)--Dongxu | * attention based QA (Solution for template expansion)--Dongxu | ||
+ | :* compute probability | ||
:* make sure whether the model is correct using the standard corpus | :* make sure whether the model is correct using the standard corpus | ||
− | + | ||
− | + | ||
* attention and memory QA(2016.2-2016.4)--Tianyi | * attention and memory QA(2016.2-2016.4)--Tianyi | ||
*Dongxu Investigation first | *Dongxu Investigation first |
2016年3月3日 (四) 01:09的最后版本
Environment
RNN Poem Processing
- update a new version -Qixin
RNN Writing Processing
- neural model learnig transform --Xingchao
- BPL speed up (transform first,then do this)
RNN QA
- attention based QA (Solution for template expansion)--Dongxu
- compute probability
- make sure whether the model is correct using the standard corpus
- attention and memory QA(2016.2-2016.4)--Tianyi
- Dongxu Investigation first
- rnn rescore--Bingdong
- rnn rescore should be used as a factor for learning to rank
- query expansion--Bingdong
Couplet
- bug fits