Language model inference
|
|
2
|
2004
|
July 27, 2017
|
Use language model in translation
|
|
0
|
1617
|
July 26, 2017
|
Teacher forcing
|
|
11
|
4138
|
July 25, 2017
|
Save validation translations at each epoch
|
|
54
|
6307
|
July 14, 2017
|
Extending the translation server to add features in source text
|
|
13
|
1634
|
July 13, 2017
|
Creating custom cost metric
|
|
4
|
2050
|
July 8, 2017
|
Local Attention
|
|
3
|
3387
|
July 8, 2017
|
Improving word feature labels by block text labeling for context
|
|
2
|
1383
|
July 3, 2017
|
Adding ROUGE to Validation metric
|
|
4
|
2493
|
June 29, 2017
|
Averaging various epochs models
|
|
11
|
2323
|
June 27, 2017
|
Adding perplexity or negative log-likelihood to CTranslate
|
|
3
|
1690
|
June 26, 2017
|
Layers config settings
|
|
0
|
964
|
June 21, 2017
|
Multi-gpu PyTorch version?
|
|
5
|
2992
|
June 6, 2017
|
OpenNMT tokenizer
|
|
1
|
1347
|
June 2, 2017
|
Weight table:set weight of some vocabulary
|
|
5
|
1508
|
May 15, 2017
|
Support other forms of dropout/regularization
|
|
8
|
1830
|
May 3, 2017
|
Resume training alternatives
|
|
14
|
2545
|
April 6, 2017
|
Minimum Risk Training for NMT
|
|
0
|
1989
|
April 3, 2017
|
Ensemble translation
|
|
2
|
1795
|
March 30, 2017
|
A w2v-coupled training procedure
|
|
0
|
1152
|
March 12, 2017
|
Model compression via weight tying
|
|
1
|
1502
|
March 6, 2017
|
Support Crayon Logging
|
|
6
|
1603
|
March 6, 2017
|
Alternate method of requesting vocab_size during preprocessing
|
|
2
|
1071
|
March 2, 2017
|
Pre-trained/fixed embeddings for features
|
|
0
|
964
|
March 2, 2017
|
Support n_best option for CTranslate
|
|
4
|
1519
|
February 22, 2017
|
It will be better if there is a Coverage Attention Module
|
|
2
|
1895
|
February 21, 2017
|
Pre-processing steps to split?
|
|
1
|
1162
|
February 20, 2017
|
Same source/target embeddings for intralingual task
|
|
2
|
1072
|
February 9, 2017
|
Creating phrase tables out of alignment output for translating UNKs
|
|
2
|
2141
|
January 16, 2017
|
External reward signal for human-in-the-loop training
|
|
2
|
1111
|
January 20, 2017
|