moses-smt/mosesdecoder
1
1
Fork 0
mirror of https://github.com/moses-smt/mosesdecoder.git synced 2024-09-20 07:42:21 +03:00
Code Issues Packages Projects Releases Wiki Activity

Merge branch 'master' of https://github.com/moses-smt/mosesdecoder

Browse Source
This commit is contained in:
Maarten van Gompel 2014-06-02 14:05:40 +02:00
commit bae4ad6253
10 changed files with 161 additions and 245 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

159
BUILD-INSTRUCTIONS.txt
View File

@ -1,158 +1,3 @@
PRELIMINARIES
Please see the Moses website on how to compile and run Moses
http://www.statmt.org/moses/?n=Development.GetStarted
Moses is primarily targeted at gcc on UNIX.
Moses requires gcc, Boost >= 1.36, and zlib including the headers that some
distributions package separately (i.e. -dev or -devel packages). Source is
available at http://boost.org .
There are several optional dependencies:
GIZA++ from http://code.google.com/p/giza-pp/ is used to align words in the parallel corpus during training.
Moses server requires xmlrpc-c with abyss-server. Source is available from
http://xmlrpc-c.sourceforge.net/.
The scripts support building ARPA format language models with SRILM or IRSTLM.
To apply models inside the decoder, you can use SRILM, IRSTLM, or KenLM. The
ARPA format is exchangable so that e.g. you can build a model with SRILM and
run the decoder with IRSTLM or KenLM.
If you want to use SRILM, you will need to download its source and build it.
The SRILM can be downloaded from
http://www.speech.sri.com/projects/srilm/download.html .
On x86_64, the default machine type is broken. Edit sbin/machine-type, find
this code
else if (`uname -m` == x86_64) then
set MACHINE_TYPE = i686
and change it to
else if (`uname -m` == x86_64) then
set MACHINE_TYPE = i686-m64
You may have to chmod +w sbin/machine-type first.
If you want to use IRSTLM, you will need to download its source and build it.
The IRSTLM can be downloaded from either the SourceForge website
http://sourceforge.net/projects/irstlm
or the official IRSTLM website
http://hlt.fbk.eu/en/irstlm
KenLM is included with Moses.
--------------------------------------------------------------------------
ADVICE ON INSTALLING EXTERNAL LIBRARIES
Generally, for trouble installing external libraries, you should get support
directly from the library maker:
Boost: http://www.boost.org/doc/libs/release/more/getting_started/unix-variants.html
IRSTLM: https://list.fbk.eu/sympa/subscribe/user-irstlm
SRILM: http://www.speech.sri.com/projects/srilm/#srilm-user
However, here's some general advice on installing software (for bash users):
#Determine where you want to install packages
PREFIX=$HOME/usr
#If your system has lib64 directories, lib64 should be used AND NOT lib
if [ -d /lib64 ]; then
LIBDIR=$PREFIX/lib64
else
LIBDIR=$PREFIX/lib
fi
#If you're installing to a non-standard path, tell programs where to find things:
export PATH=$PREFIX/bin${PATH:+:$PATH}
export LD_LIBRARY_PATH=$LIBDIR${LD_LIBRARY_PATH:+:$LD_LIBRARY_PATH}
export LIBRARY_PATH=$LIBDIR${LIBRARY_PATH:+:$LIBRARY_PATH}
export CPATH=$PREFIX/include${CPATH:+:$CPATH}
Add all the above code to your .bashrc or .bash_login as appropriate. Then
you're ready to install packages in non-standard paths:
#For autotools packages e.g. xmlrpc-c and zlib
./configure --prefix=$PREFIX --libdir=$LIBDIR [other options here]
#tcmalloc is a malloc implementation with threaded performance. To see how it
#improves Moses performance, read
# http://www.mail-archive.com/moses-support@mit.edu/msg07303.html
#It is part of gperftools which can be downloaded from from
# https://code.google.com/p/gperftools/downloads/list
#configure with this:
./configure --prefix=$PREFIX --libdir=$LIBDIR --enable-shared --enable-static --enable-minimal
#For bzip2:
wget http://www.bzip.org/1.0.6/bzip2-1.0.6.tar.gz
tar xzvf bzip2-1.0.6.tar.gz
cd bzip2-1.0.6/
#Compile and install libbz2.a (static library)
make
make install PREFIX=$PREFIX
mkdir -p $LIBDIR
#Note this may be the same file; you can ignore the error
mv $PREFIX/lib/libbz2.a $LIBDIR 2>/dev/null
#Compile and install libbz2.so (dynamic library)
make clean
make -f Makefile-libbz2_so
cp libbz2.so.* $LIBDIR
ln -sf libbz2.so.1.0 $LIBDIR/libbz2.so
#For Boost:
./bootstrap.sh
./b2 --prefix=$PWD --libdir=$PWD/lib64 --layout=tagged link=static,shared threading=multi,single install || echo FAILURE
This will put the header files and libraries files in the current directory, rather than the system directory.
For most Linux systems, you should replace
link=static,shared
with
link=static
so it will only create static libraries. The minimised headaches when linking with Moses.
To link Moses to your version of boost,
./bjam --with-boost=[boost/path]
Alternatively, you run
./b2 --prefix=/usr/ --libdir=/usr/lib
to install boost in the systems folder. However, this may override the built in boost and causes problems for your OS, therefore, it is not recommended.
--------------------------------------------------------------------------
BUILDING
Building consists of running
./bjam [options]
Common options are:
--with-srilm=/path/to/srilm to compile the decoder with SRILM support
--with-irstlm=/path/to/irstlm to compile the decoder with IRSTLM support
-jN where N is the number of CPUs
--with-macports=/path/to/macports use MacPorts on Mac OS X.
If you leave out /path/to/macports bjam will use the /opt/local as default.
You don't have to use --with-boost with-macports as it is implicitly set.
Also note that using --with-macports automatically triggers "using darwin".
Binaries will appear in dist/bin.
You can clean up data from previous builds using
./bjam --clean
For further documentation, run
./bjam --help
--------------------------------------------------------------------------
ALTERNATIVE WAYS TO BUILD ON UNIX AND OTHER PLATFORMS
Microsoft Windows
-----------------
Moses is primarily targeted at gcc on UNIX. Windows users should
install using Cygwin. Outdated instructions can be found here:
http://ssli.ee.washington.edu/people/amittai/Moses-on-Win7.pdf .
Binaries for all external libraries needed can be downloaded from
http://www.statmt.org/moses/?n=Moses.LibrariesUsed
Only the decoder is developed and tested under Windows. There are
difficulties using the training scripts under Windows, even with
Cygwin, but it can be done.

8
contrib/server/mosesserver.cpp
View File

@ -279,6 +279,12 @@ public:
manager.ProcessSentence();
const ChartHypothesis *hypo = manager.GetBestHypothesis();
outputChartHypo(out,hypo);
if (addGraphInfo) {
const size_t translationId = tinput.GetTranslationId();
std::ostringstream sgstream;
manager.GetSearchGraph(translationId,sgstream);
retData.insert(pair<string, xmlrpc_c::value>("sg", xmlrpc_c::value_string(sgstream.str())));
}
} else {
Sentence sentence;
const vector<FactorType> &inputFactorOrder =
@ -310,7 +316,7 @@ public:
retData.insert(pair<string, xmlrpc_c::value_array>("word-align", alignments));
}
if(addGraphInfo) {
if (addGraphInfo) {
insertGraphInfo(manager,retData);
(const_cast<StaticData&>(staticData)).SetOutputSearchGraph(false);
}

123
moses/FF/ConstrainedDecoding.cpp
View File

@ -25,7 +25,7 @@ ConstrainedDecodingState::ConstrainedDecodingState(const ChartHypothesis &hypo)
int ConstrainedDecodingState::Compare(const FFState& other) const
{
const ConstrainedDecodingState &otherFF = static_cast<const ConstrainedDecodingState&>(other);
int ret = m_outputPhrase.Compare(otherFF.m_outputPhrase);
int ret = m_outputPhrase.Compare(otherFF.m_outputPhrase);
return ret;
}
@ -34,6 +34,7 @@ ConstrainedDecoding::ConstrainedDecoding(const std::string &line)
:StatefulFeatureFunction(1, line)
,m_maxUnknowns(0)
,m_negate(false)
,m_soft(false)
{
m_tuneable = false;
ReadParameters();
@ -44,47 +45,48 @@ void ConstrainedDecoding::Load()
const StaticData &staticData = StaticData::Instance();
bool addBeginEndWord = (staticData.GetSearchAlgorithm() == ChartDecoding) || (staticData.GetSearchAlgorithm() == ChartIncremental);
InputFileStream constraintFile(m_path);
std::string line;
long sentenceID = staticData.GetStartTranslationId() - 1;
while (getline(constraintFile, line)) {
vector<string> vecStr = Tokenize(line, "\t");
Phrase phrase(0);
if (vecStr.size() == 1) {
sentenceID++;
phrase.CreateFromString(Output, staticData.GetOutputFactorOrder(), vecStr[0], staticData.GetFactorDelimiter(), NULL);
} else if (vecStr.size() == 2) {
sentenceID = Scan<long>(vecStr[0]);
phrase.CreateFromString(Output, staticData.GetOutputFactorOrder(), vecStr[1], staticData.GetFactorDelimiter(), NULL);
} else {
UTIL_THROW(util::Exception, "Reference file not loaded");
for(size_t i = 0; i < m_paths.size(); ++i) {
InputFileStream constraintFile(m_paths[i]);
std::string line;
long sentenceID = staticData.GetStartTranslationId() - 1;
while (getline(constraintFile, line)) {
vector<string> vecStr = Tokenize(line, "\t");
Phrase phrase(0);
if (vecStr.size() == 1) {
sentenceID++;
phrase.CreateFromString(Output, staticData.GetOutputFactorOrder(), vecStr[0], staticData.GetFactorDelimiter(), NULL);
} else if (vecStr.size() == 2) {
sentenceID = Scan<long>(vecStr[0]);
phrase.CreateFromString(Output, staticData.GetOutputFactorOrder(), vecStr[1], staticData.GetFactorDelimiter(), NULL);
} else {
UTIL_THROW(util::Exception, "Reference file not loaded");
}
if (addBeginEndWord) {
phrase.InitStartEndWord();
}
m_constraints[sentenceID].push_back(phrase);
}
if (addBeginEndWord) {
phrase.InitStartEndWord();
}
m_constraints.insert(make_pair(sentenceID,phrase));
}
}
std::vector<float> ConstrainedDecoding::DefaultWeights() const
{
UTIL_THROW_IF2(m_numScoreComponents != 1,
"ConstrainedDecoding must only have 1 score");
"ConstrainedDecoding must only have 1 score");
vector<float> ret(1, 1);
return ret;
}
template <class H, class M>
const Phrase *GetConstraint(const std::map<long,Phrase> &constraints, const H &hypo)
const std::vector<Phrase> *GetConstraint(const std::map<long,std::vector<Phrase> > &constraints, const H &hypo)
{
const M &mgr = hypo.GetManager();
const InputType &input = mgr.GetSource();
long id = input.GetTranslationId();
map<long,Phrase>::const_iterator iter;
map<long,std::vector<Phrase> >::const_iterator iter;
iter = constraints.find(id);
if (iter == constraints.end()) {
@ -101,30 +103,37 @@ FFState* ConstrainedDecoding::Evaluate(
const FFState* prev_state,
ScoreComponentCollection* accumulator) const
{
const Phrase *ref = GetConstraint<Hypothesis, Manager>(m_constraints, hypo);
const std::vector<Phrase> *ref = GetConstraint<Hypothesis, Manager>(m_constraints, hypo);
assert(ref);
ConstrainedDecodingState *ret = new ConstrainedDecodingState(hypo);
const Phrase &outputPhrase = ret->GetPhrase();
const Phrase &outputPhrase = ret->GetPhrase();
size_t searchPos = ref->Find(outputPhrase, m_maxUnknowns);
size_t searchPos = NOT_FOUND;
size_t i = 0;
size_t size = 0;
while(searchPos == NOT_FOUND && i < ref->size()) {
searchPos = (*ref)[i].Find(outputPhrase, m_maxUnknowns);
size = (*ref)[i].GetSize();
i++;
}
float score;
if (hypo.IsSourceCompleted()) {
// translated entire sentence.
bool match = (searchPos == 0) && (ref->GetSize() == outputPhrase.GetSize());
if (!m_negate) {
score = match ? 0 : - std::numeric_limits<float>::infinity();
}
else {
score = !match ? 0 : - std::numeric_limits<float>::infinity();
}
bool match = (searchPos == 0) && (size == outputPhrase.GetSize());
if (!m_negate) {
score = match ? 0 : - ( m_soft ? 1 : std::numeric_limits<float>::infinity());
}
else {
score = !match ? 0 : - ( m_soft ? 1 : std::numeric_limits<float>::infinity());
}
} else if (m_negate) {
// keep all derivations
score = 0;
// keep all derivations
score = 0;
}
else {
score = (searchPos != NOT_FOUND) ? 0 : - std::numeric_limits<float>::infinity();
score = (searchPos != NOT_FOUND) ? 0 : - ( m_soft ? 1 : std::numeric_limits<float>::infinity());
}
accumulator->PlusEquals(this, score);
@ -137,7 +146,7 @@ FFState* ConstrainedDecoding::EvaluateChart(
int /* featureID - used to index the state in the previous hypotheses */,
ScoreComponentCollection* accumulator) const
{
const Phrase *ref = GetConstraint<ChartHypothesis, ChartManager>(m_constraints, hypo);
const std::vector<Phrase> *ref = GetConstraint<ChartHypothesis, ChartManager>(m_constraints, hypo);
assert(ref);
const ChartManager &mgr = hypo.GetManager();
@ -145,25 +154,33 @@ FFState* ConstrainedDecoding::EvaluateChart(
ConstrainedDecodingState *ret = new ConstrainedDecodingState(hypo);
const Phrase &outputPhrase = ret->GetPhrase();
size_t searchPos = ref->Find(outputPhrase, m_maxUnknowns);
size_t searchPos = NOT_FOUND;
size_t i = 0;
size_t size = 0;
while(searchPos == NOT_FOUND && i < ref->size()) {
searchPos = (*ref)[i].Find(outputPhrase, m_maxUnknowns);
size = (*ref)[i].GetSize();
i++;
}
float score;
if (hypo.GetCurrSourceRange().GetStartPos() == 0 &&
hypo.GetCurrSourceRange().GetEndPos() == source.GetSize() - 1) {
// translated entire sentence.
bool match = (searchPos == 0) && (ref->GetSize() == outputPhrase.GetSize());
bool match = (searchPos == 0) && (size == outputPhrase.GetSize());
if (!m_negate) {
score = match ? 0 : - std::numeric_limits<float>::infinity();
}
else {
score = !match ? 0 : - std::numeric_limits<float>::infinity();
}
if (!m_negate) {
score = match ? 0 : - ( m_soft ? 1 : std::numeric_limits<float>::infinity());
}
else {
score = !match ? 0 : - ( m_soft ? 1 : std::numeric_limits<float>::infinity());
}
} else if (m_negate) {
// keep all derivations
score = 0;
// keep all derivations
score = 0;
} else {
score = (searchPos != NOT_FOUND) ? 0 : - std::numeric_limits<float>::infinity();
score = (searchPos != NOT_FOUND) ? 0 : - ( m_soft ? 1 : std::numeric_limits<float>::infinity());
}
accumulator->PlusEquals(this, score);
@ -174,11 +191,13 @@ FFState* ConstrainedDecoding::EvaluateChart(
void ConstrainedDecoding::SetParameter(const std::string& key, const std::string& value)
{
if (key == "path") {
m_path = value;
m_paths = Tokenize(value, ",");
} else if (key == "max-unknowns") {
m_maxUnknowns = Scan<int>(value);
} else if (key == "negate") {
m_negate = Scan<bool>(value);
m_negate = Scan<bool>(value);
} else if (key == "soft") {
m_soft = Scan<bool>(value);
} else {
StatefulFeatureFunction::SetParameter(key, value);
}

9
moses/FF/ConstrainedDecoding.h
View File

@ -46,13 +46,15 @@ public:
, ScoreComponentCollection &scoreBreakdown
, ScoreComponentCollection &estimatedFutureScore) const
{}
void Evaluate(const InputType &input
void Evaluate(const InputType &input
, const InputPath &inputPath
, const TargetPhrase &targetPhrase
, const StackVec *stackVec
, ScoreComponentCollection &scoreBreakdown
, ScoreComponentCollection *estimatedFutureScore = NULL) const
{}
FFState* Evaluate(
const Hypothesis& cur_hypo,
const FFState* prev_state,
@ -72,10 +74,11 @@ public:
void SetParameter(const std::string& key, const std::string& value);
protected:
std::string m_path;
std::map<long,Phrase> m_constraints;
std::vector<std::string> m_paths;
std::map<long, std::vector<Phrase> > m_constraints;
int m_maxUnknowns;
bool m_negate; // only keep translations which DON'T match the reference
bool m_soft;
};

46
moses/FF/NieceTerminal.cpp
View File

@ -10,6 +10,13 @@ using namespace std;
namespace Moses
{
NieceTerminal::NieceTerminal(const std::string &line)
:StatelessFeatureFunction(line)
,m_hardConstraint(false)
{
ReadParameters();
}
void NieceTerminal::Evaluate(const Phrase &source
, const TargetPhrase &targetPhrase
, ScoreComponentCollection &scoreBreakdown
@ -25,6 +32,8 @@ void NieceTerminal::Evaluate(const InputType &input
, ScoreComponentCollection &scoreBreakdown
, ScoreComponentCollection *estimatedFutureScore) const
{
assert(stackVec);
const Phrase *ruleSource = targetPhrase.GetRuleSource();
assert(ruleSource);
@ -36,23 +45,19 @@ void NieceTerminal::Evaluate(const InputType &input
}
}
size_t ntInd = 0;
for (size_t i = 0; i < ruleSource->GetSize(); ++i) {
const Word &word = ruleSource->GetWord(i);
if (word.IsNonTerminal()) {
const ChartCellLabel &cell = *stackVec->at(ntInd);
const WordsRange &ntRange = cell.GetCoverage();
bool containTerm = ContainTerm(input, ntRange, terms);
for (size_t i = 0; i < stackVec->size(); ++i) {
const ChartCellLabel &cell = *stackVec->at(i);
const WordsRange &ntRange = cell.GetCoverage();
bool containTerm = ContainTerm(input, ntRange, terms);
if (containTerm) {
//cerr << "ruleSource=" << *ruleSource << " ";
//cerr << "ntRange=" << ntRange << endl;
if (containTerm) {
//cerr << "ruleSource=" << *ruleSource << " ";
//cerr << "ntRange=" << ntRange << endl;
// non-term contains 1 of the terms in the rule.
scoreBreakdown.PlusEquals(this, 1);
return;
}
++ntInd;
// non-term contains 1 of the terms in the rule.
float score = m_hardConstraint ? - std::numeric_limits<float>::infinity() : 1;
scoreBreakdown.PlusEquals(this, score);
return;
}
}
@ -83,5 +88,16 @@ bool NieceTerminal::ContainTerm(const InputType &input,
return false;
}
void NieceTerminal::SetParameter(const std::string& key, const std::string& value)
{
if (key == "hard-constraint") {
m_hardConstraint = Scan<bool>(value);
} else {
StatelessFeatureFunction::SetParameter(key, value);
}
}
}

8
moses/FF/NieceTerminal.h
View File

@ -13,9 +13,7 @@ class Word;
class NieceTerminal : public StatelessFeatureFunction
{
public:
NieceTerminal(const std::string &line)
:StatelessFeatureFunction(line)
{}
NieceTerminal(const std::string &line);
bool IsUseable(const FactorMask &mask) const {
return true;
@ -36,7 +34,10 @@ public:
void EvaluateChart(const ChartHypothesis &hypo,
ScoreComponentCollection* accumulator) const;
void SetParameter(const std::string& key, const std::string& value);
protected:
bool m_hardConstraint;
bool ContainTerm(const InputType &input,
const WordsRange &ntRange,
const std::set<Word> &terms) const;
@ -44,3 +45,4 @@ protected:
}

33
moses/Parameter.cpp
View File

@ -191,6 +191,7 @@ Parameter::Parameter()
AddParam("weight", "weights for ALL models, 1 per line 'WeightName value'. Weight names can be repeated");
AddParam("weight-overwrite", "special parameter for mert. All on 1 line. Overrides weights specified in 'weights' argument");
AddParam("feature-overwrite", "Override arguments in a particular feature function with a particular key. Format: -feature-overwrite \"FeatureName key=value\"");
AddParam("weight-add", "Add weight for FF if it doesn't exist, i.e weights here are added 1st, and can be override by the ini file or on the command line. Used to specify initial weights for FF that was also specified on the copmmand line");
AddParam("feature-add", "Add a feature function on the command line. Used by mira to add BLEU feature");
AddParam("feature-name-overwrite", "Override feature name (NOT arguments). Eg. SRILM-->KENLM, PhraseDictionaryMemory-->PhraseDictionaryScope3");
@ -903,24 +904,28 @@ void Parameter::ConvertWeightArgs()
void Parameter::CreateWeightsMap()
{
PARAM_VEC &vec = m_setting["weight"];
CreateWeightsMap(m_setting["weight-add"]);
CreateWeightsMap(m_setting["weight"]);
}
void Parameter::CreateWeightsMap(const PARAM_VEC &vec)
{
for (size_t i = 0; i < vec.size(); ++i) {
const string &line = vec[i];
vector<string> toks = Tokenize(line);
UTIL_THROW_IF2(toks.size() < 2,
"Error in format of weights: " << line);
const string &line = vec[i];
vector<string> toks = Tokenize(line);
UTIL_THROW_IF2(toks.size() < 2,
"Error in format of weights: " << line);
string name = toks[0];
name = name.substr(0, name.size() - 1);
string name = toks[0];
name = name.substr(0, name.size() - 1);
vector<float> weights(toks.size() - 1);
for (size_t i = 1; i < toks.size(); ++i) {
float weight = Scan<float>(toks[i]);
weights[i - 1] = weight;
}
m_weights[name] = weights;
vector<float> weights(toks.size() - 1);
for (size_t i = 1; i < toks.size(); ++i) {
float weight = Scan<float>(toks[i]);
weights[i - 1] = weight;
}
m_weights[name] = weights;
}
}
void Parameter::WeightOverwrite()

1
moses/Parameter.h
View File

@ -76,6 +76,7 @@ protected:
void ConvertWeightArgsWordPenalty();
void ConvertPhrasePenalty();
void CreateWeightsMap();
void CreateWeightsMap(const PARAM_VEC &vec);
void WeightOverwrite();
void AddFeature(const std::string &line);
void AddFeaturesCmd();

1
scripts/ems/experiment.perl
View File

@ -633,6 +633,7 @@ sub check_producability {
return 1 if defined($CONFIG{$out});
# find defined step that produces this
$out =~ s/:.+:/:/g;
my $defined_step;
foreach my $ds (keys %STEP_OUT) {
my ($ds_module) = &deconstruct_name($ds);

18
scripts/tokenizer/remove-non-printing-char.perl Executable file
View File

@ -0,0 +1,18 @@
#!/usr/bin/perl
use utf8;
binmode(STDIN, ":utf8");
binmode(STDOUT, ":utf8");
binmode(STDERR, ":utf8");
while (my $line = <STDIN>) {
chomp($line);
#$line =~ tr/\040-\176/ /c;
#$line =~ s/[^[:print:]]/ /g;
#$line =~ s/\s+/ /g;
$line =~ s/\p{C}/ /g;
print "$line\n";
}