“Tianyi Luo 2016-05-02”版本间的差异
来自cslt Wiki
（以“=== Plan to do this week === * To implement tensorflow version of RNN/LSTM Max margin vector training. === Work done in this week === --------------------2016-04-26...”为内容创建页面） |
|||
| （相同用户的6个中间修订版本未显示） | |||
| 第1行： | 第1行： | ||
=== Plan to do this week === | === Plan to do this week === | ||
* To implement tensorflow version of RNN/LSTM Max margin vector training. | * To implement tensorflow version of RNN/LSTM Max margin vector training. | ||
| + | * To implement attention chatting model with xiaobing corpus. | ||
=== Work done in this week === | === Work done in this week === | ||
| + | --------------------2016-04-25 | ||
| + | * Conduct some preprocessing of xiaobing corpus. | ||
--------------------2016-04-26 | --------------------2016-04-26 | ||
| − | + | * Help jiyuan to understand the music generation paper. | |
| + | * Check jiyuan's code(from RNN-RBM to LSTM-RBM). | ||
--------------------2016-04-27 | --------------------2016-04-27 | ||
| − | + | * Modify the LSTM Max margin vector training code. Its cost is lower than RNN and the training speed is slower than RNN. The performance improves from 82.39% to 84.83%. | |
--------------------2016-04-28 | --------------------2016-04-28 | ||
| − | + | * Finish part of code about qaa(Current sample is q1; Positive sample is a1; Negtive sampling is a2.) max margin theano version. | |
--------------------2016-04-29 | --------------------2016-04-29 | ||
| − | + | * Finish the code about qaa(Current sample is q1; Positive sample is a1; Negtive sampling is a2.) max margin theano version. | |
| − | --------------------2016-04-30 | + | --------------------2016-04-30~2016-05-01 |
| − | + | The Holiday of Labors Day | |
| − | + | ||
| − | + | ||
| − | + | ||
2016年5月9日 (一) 10:44的最后版本
Plan to do this week
- To implement tensorflow version of RNN/LSTM Max margin vector training.
- To implement attention chatting model with xiaobing corpus.
Work done in this week
2016-04-25
- Conduct some preprocessing of xiaobing corpus.
2016-04-26
- Help jiyuan to understand the music generation paper.
- Check jiyuan's code(from RNN-RBM to LSTM-RBM).
2016-04-27
- Modify the LSTM Max margin vector training code. Its cost is lower than RNN and the training speed is slower than RNN. The performance improves from 82.39% to 84.83%.
2016-04-28
- Finish part of code about qaa(Current sample is q1; Positive sample is a1; Negtive sampling is a2.) max margin theano version.
2016-04-29
- Finish the code about qaa(Current sample is q1; Positive sample is a1; Negtive sampling is a2.) max margin theano version.
2016-04-30~2016-05-01
The Holiday of Labors Day
Plan to do next week
- To implement tensorflow version of RNN/LSTM Max margin vector training.
- To implement attention chatting model with xiaobing corpus.
Interested papers
- Cascading Bandits: Learning to Rank in the Cascade Model(ICML 2015) [pdf]
