Sparse features trainable with hg mira

(In phrase-based Moses)

6 files changed, 15 insertions, 45 deletions

diff --git a/moses/FeatureVector.cpp b/moses/FeatureVector.cpp
index 536c1a720..fd44b1002 100644
--- a/moses/FeatureVector.cpp
+++ b/moses/FeatureVector.cpp
@@ -214,10 +214,10 @@ void FVector::save(const string& filename) const
   out.close();
 }
 
-void FVector::write(ostream& out) const
+void FVector::write(ostream& out,const string& sep) const
 {
   for (const_iterator i = cbegin(); i != cend(); ++i) {
-    out << i->first << " " << i->second << endl;
+    out << i->first << sep << i->second << endl;
   }
 }
 
diff --git a/moses/FeatureVector.h b/moses/FeatureVector.h
index 65f8bdc2e..e14006da7 100644
--- a/moses/FeatureVector.h
+++ b/moses/FeatureVector.h
@@ -167,7 +167,7 @@ public:
   /** Load from file - each line should be 'root[_name] value' */
   bool load(const std::string& filename);
   void save(const std::string& filename) const;
-  void write(std::ostream& out) const ;
+  void write(std::ostream& out, const std::string& sep=" ") const ;
 
   /** Element access */
   ProxyFVector operator[](const FName& name);
diff --git a/moses/Manager.cpp b/moses/Manager.cpp
index 196f4d997..3cffa8503 100644
--- a/moses/Manager.cpp
+++ b/moses/Manager.cpp
@@ -757,18 +757,7 @@ void Manager::OutputFeatureValuesForHypergraph(const Hypothesis* hypo, std::ostr
 {
   outputSearchGraphStream.setf(std::ios::fixed);
   outputSearchGraphStream.precision(6);
-
-  const vector<const StatelessFeatureFunction*>& slf =StatelessFeatureFunction::GetStatelessFeatureFunctions();
-  const vector<const StatefulFeatureFunction*>& sff = StatefulFeatureFunction::GetStatefulFeatureFunctions();
-  size_t featureIndex = 1;
-  for (size_t i = 0; i < sff.size(); ++i) {
-    featureIndex = OutputFeatureValuesForHypergraph(featureIndex, hypo, sff[i], outputSearchGraphStream);
-  }
-  for (size_t i = 0; i < slf.size(); ++i) {
-    {
-      featureIndex = OutputFeatureValuesForHypergraph(featureIndex, hypo, slf[i], outputSearchGraphStream);
-    }
-  }
+  hypo->GetScoreBreakdown().Save(outputSearchGraphStream, false);
 }
 
 
@@ -833,30 +822,6 @@ size_t Manager::OutputFeatureValuesForSLF(size_t index, bool zeros, const Hypoth
   // }
 }
 
-size_t Manager::OutputFeatureValuesForHypergraph(size_t index, const Hypothesis* hypo, const FeatureFunction* ff, std::ostream &outputSearchGraphStream) const
-{
-  if (!ff->IsTuneable()) {
-    return index;
-  }
-  ScoreComponentCollection scoreCollection = hypo->GetScoreBreakdown();
-  const Hypothesis *prevHypo = hypo->GetPrevHypo();
-  if (prevHypo) {
-    scoreCollection.MinusEquals( prevHypo->GetScoreBreakdown() );
-  }
-  vector<float> featureValues = scoreCollection.GetScoresForProducer(ff);
-  size_t numScoreComps = featureValues.size();
-
-  if (numScoreComps > 1) {
-    for (size_t i = 0; i < numScoreComps; ++i) {
-      outputSearchGraphStream << ff->GetScoreProducerDescription()  << i << "=" << featureValues[i] << " ";
-    }
-  } else {
-    outputSearchGraphStream << ff->GetScoreProducerDescription()  << "=" << featureValues[0] << " ";
-  }
-
-  return index+numScoreComps;
-}
-
 /**! Output search graph in hypergraph format of Kenneth Heafield's lazy hypergraph decoder */
 void Manager::OutputSearchGraphAsHypergraph(long translationId, std::ostream &outputSearchGraphStream) const
 {
diff --git a/moses/Manager.h b/moses/Manager.h
index ccf57c527..c015b701c 100644
--- a/moses/Manager.h
+++ b/moses/Manager.h
@@ -106,7 +106,6 @@ private:
 
   // Helper functions to output search graph in the hypergraph format of Kenneth Heafield's lazy hypergraph decoder
   void OutputFeatureValuesForHypergraph(const Hypothesis* hypo, std::ostream &outputSearchGraphStream) const;
-  size_t OutputFeatureValuesForHypergraph(size_t index, const Hypothesis* hypo, const FeatureFunction* ff, std::ostream &outputSearchGraphStream) const;
 
 
 protected:
diff --git a/moses/ScoreComponentCollection.cpp b/moses/ScoreComponentCollection.cpp
index 52ec00dd4..d4df4742e 100644
--- a/moses/ScoreComponentCollection.cpp
+++ b/moses/ScoreComponentCollection.cpp
@@ -179,25 +179,31 @@ void ScoreComponentCollection::SparseL2Regularize(float lambda)
   m_scores.sparseL2regularize(lambda);
 }
 
-void ScoreComponentCollection::Save(ostream& out) const
+void ScoreComponentCollection::Save(ostream& out, bool multiline) const
 {
+  string sep = " ";
+  string linesep = "\n";
+  if (!multiline) {
+     sep = "=";
+     linesep = " ";
+  }
   ScoreIndexMap::const_iterator iter = s_scoreIndexes.begin();
   for (; iter != s_scoreIndexes.end(); ++iter ) {
     string name = iter->first->GetScoreProducerDescription();
     IndexPair ip = iter->second; // feature indices
     if (ip.second-ip.first == 1) {
-      out << name << " " << m_scores[ip.first] << endl;
+      out << name << sep << m_scores[ip.first] << linesep;
     } else {
       for (size_t i=ip.first; i < ip.second; ++i) {
         ostringstream fullname;
         fullname << name << "_" << (i + 1 - ip.first);
-        out << fullname.str() << " " << m_scores[i] << endl;
+        out << fullname.str() << sep << m_scores[i] << linesep;
       }
     }
   }
 
   // write sparse features
-  m_scores.write(out);
+  m_scores.write(out,sep);
 }
 
 void ScoreComponentCollection::Save(const string& filename) const
diff --git a/moses/ScoreComponentCollection.h b/moses/ScoreComponentCollection.h
index a7fa65883..858fcd5a8 100644
--- a/moses/ScoreComponentCollection.h
+++ b/moses/ScoreComponentCollection.h
@@ -383,7 +383,7 @@ public:
   size_t SparseL1Regularize(float lambda);
   void SparseL2Regularize(float lambda);
   void Save(const std::string& filename) const;
-  void Save(std::ostream&) const;
+  void Save(std::ostream&, bool multiline=true) const;
 
   void IncrementSparseHopeFeatures() {
     m_scores.incrementSparseHopeFeatures();