“14-10-19 Dongxu Zhang”版本间的差异
来自cslt Wiki
（以“Last week ---------------------- 1. Train LSTM-Rnn LM with 200MB corpus(vocabulary 10k, classes 100). when using 2 kernels, it takes aroung 200min per epoch. 2. Trai...”为内容创建页面） |
|||
| 第1行： | 第1行： | ||
| − | + | === Accomplished this week === | |
| − | + | * 1. Train LSTM-Rnn LM with 200MB corpus(vocabulary 10k, classes 100). when using 2 kernels, it takes aroung 200min per epoch. | |
| − | 1. Train LSTM-Rnn LM with 200MB corpus(vocabulary 10k, classes 100). when using 2 kernels, it takes aroung 200min per epoch. | + | * 2. Train 5-gram LM using Baiduzhidao_corpus(~30GB after preprocess) with new lexicon. There is a mistake when counted possiblity after merge. |
| − | 2. Train 5-gram LM using Baiduzhidao_corpus(~30GB after preprocess) with new lexicon. There is a mistake when counted possiblity after merge. | + | * 3. An idea occured to me which may improve word2vec with much more semantic information. But there is huge computation complexity problem that bothers me, which I wish we can discuss. |
| − | 3. An idea occured to me which may improve word2vec with much more semantic information. But there is huge computation complexity problem that bothers me, which I wish we can discuss. | + | * 4. Read paper "Learning Long-Term Dependencies with Gradient Descent is Difficult". Still in progress. |
| − | 4. Read paper "Learning Long-Term Dependencies with Gradient Descent is Difficult". Still in progress. | + | |
| − | + | === Next week === | |
| − | + | * 1. Test LSTM-Rnn LM. | |
| − | 1. Test LSTM-Rnn LM. | + | * 2. Finished building lexion. |
| − | 2. Finished building lexion. | + | * 3. Understand the paper. |
| − | 3. Understand the paper. | + | * 4. May have time to achieve my baseline idea on text8. |
| − | 4. May have time to achieve my baseline idea on text8. | + | |
2014年10月19日 (日) 12:42的版本
Accomplished this week
- 1. Train LSTM-Rnn LM with 200MB corpus(vocabulary 10k, classes 100). when using 2 kernels, it takes aroung 200min per epoch.
- 2. Train 5-gram LM using Baiduzhidao_corpus(~30GB after preprocess) with new lexicon. There is a mistake when counted possiblity after merge.
- 3. An idea occured to me which may improve word2vec with much more semantic information. But there is huge computation complexity problem that bothers me, which I wish we can discuss.
- 4. Read paper "Learning Long-Term Dependencies with Gradient Descent is Difficult". Still in progress.
Next week
- 1. Test LSTM-Rnn LM.
- 2. Finished building lexion.
- 3. Understand the paper.
- 4. May have time to achieve my baseline idea on text8.
