Commits

Commits on Jun 23, 2016

clean up loadVocab.m
lmthang
committedJun 23, 2016

Commits on Jun 7, 2016

decoder ensures that the first symbol is not eos
lmthang
committedJun 7, 2016

Commits on May 22, 2016

update comments
lmthang
committedMay 22, 2016

Commits on May 20, 2016

try a different way of normalizing local attention
lmthang
committedMay 20, 2016
update grad_checks.txt
lmthang
committedMay 20, 2016
fix an important bug for normLocalAttn backprop
lmthang
committedMay 20, 2016

Commits on May 18, 2016

normalize local attention
lmthang
committedMay 18, 2016

Commits on Apr 22, 2016

addNoise option to add some pertubation to loaded weights
lmthang
committedApr 22, 2016

Commits on Apr 19, 2016

Merge branch 'master' of ssh:https://jamie.stanford.edu/scr/nlp/deeplearning/deepmt/git/nmt.matlab
lmthang
committedApr 19, 2016
minor fix for decoding local attention, set srcMaxLen
lmthang
committedApr 19, 2016

Commits on Mar 29, 2016

Merge branch 'master' of /scr/nlp/deeplearning/deepmt/git/nmt.matlab
lmthang
committedMar 29, 2016
add chrF.py, update disp_translations.py for analysis
lmthang
committedMar 29, 2016
fix decoder to replicate srcMask for attention models
lmthang
committedMar 29, 2016

Commits on Mar 27, 2016

remove totalScore in computeSentRepresentations.m
lmthang
committedMar 27, 2016
add runEncoder.m computeSentRepresentations.m
lmthang
committedMar 27, 2016
factor encoder code in lstmDecoder.m, add computeSentRepresentations.m
lmthang
committedMar 27, 2016

Commits on Mar 17, 2016

Merge branch 'master' of ssh:https://jamie.stanford.edu/scr/nlp/deeplearning/deepmt/git/nmt.matlab
lmthang
committedMar 17, 2016
add stopId to testLSTM
lmthang
committedMar 17, 2016

Commits on Mar 16, 2016

Merge branch 'master' of /scr/nlp/deeplearning/deepmt/git/nmt.matlab
lmthang
committedMar 16, 2016
fix preprocessing steps, also minor fix to getAlignmentWeights to handle non-attention model
lmthang
committedMar 16, 2016
optimize lstmUnitForward.m, good speedup:)
lmthang
committedMar 16, 2016

Commits on Mar 7, 2016

fix global attention to zero out alignments at masked locations
lmthang
committedMar 7, 2016

Commits on Mar 6, 2016

minor fix for computeRerankScores to run on GPU
lmthang
committedMar 6, 2016
update printScore
lmthang
committedMar 6, 2016
Merge branch 'master' of ssh:https://jamie.stanford.edu/scr/nlp/deeplearning/deepmt/git/nmt.matlab
lmthang
committedMar 6, 2016
add computeRerankScores
lmthang
committedMar 6, 2016

Commits on Mar 5, 2016

Commits on Feb 17, 2016

Merge branch 'master' of /scr/nlp/deeplearning/deepmt/git/nmt.matlab
lmthang
committedFeb 17, 2016
remove assert(bestWords == params.tgtEos)
lmthang
committedFeb 17, 2016
add continueId to testLSTM to resume decoding
lmthang
committedFeb 17, 2016

Commits on Feb 16, 2016

minor updates to trainLSTM for changes in the decoder
lmthang
committedFeb 16, 2016
done adding prefix decoder, clean up force decoder, allow to print decoding scores
lmthang
committedFeb 16, 2016