Move training routine out of ff_const_reorder_common.h

5 files changed, 91 insertions, 8 deletions

diff --git a/training/const_reorder/Makefile.am b/training/const_reorder/Makefile.am
index 2e81e588..367ac904 100644
--- a/training/const_reorder/Makefile.am
+++ b/training/const_reorder/Makefile.am
@@ -1,8 +1,12 @@
+noinst_LIBRARIES = libtrainer.a
+
+libtrainer_a_SOURCES = trainer.h trainer.cc
+
 bin_PROGRAMS = const_reorder_model_trainer argument_reorder_model_trainer
 
 AM_CPPFLAGS = -I$(top_srcdir) -I$(top_srcdir)/utils -I$(top_srcdir)/decoder
 
 const_reorder_model_trainer_SOURCES = constituent_reorder_model.cc
-const_reorder_model_trainer_LDADD = ../../utils/libutils.a
+const_reorder_model_trainer_LDADD = ../../utils/libutils.a libtrainer.a
 argument_reorder_model_trainer_SOURCES = argument_reorder_model.cc
-argument_reorder_model_trainer_LDADD = ../../utils/libutils.a
+argument_reorder_model_trainer_LDADD = ../../utils/libutils.a libtrainer.a
diff --git a/training/const_reorder/argument_reorder_model.cc b/training/const_reorder/argument_reorder_model.cc
index 54402436..87f2ce2f 100644
--- a/training/const_reorder/argument_reorder_model.cc
+++ b/training/const_reorder/argument_reorder_model.cc
@@ -14,7 +14,7 @@
 
 #include "utils/filelib.h"
 
-#include "decoder/ff_const_reorder_common.h"
+#include "trainer.h"
 
 using namespace std;
 using namespace const_reorder;
@@ -93,8 +93,8 @@ struct SArgumentReorderTrainer {
       strcpy(pszNewInstanceFName, pszInstanceFname);
     }
 
-    Tsuruoka_Maxent* pMaxent = new Tsuruoka_Maxent(NULL);
-    pMaxent->fnTrain(pszNewInstanceFName, "l1", pszModelFname, 300);
+    Tsuruoka_Maxent_Trainer* pMaxent = new Tsuruoka_Maxent_Trainer;
+    pMaxent->fnTrain(pszNewInstanceFName, "l1", pszModelFname);
     delete pMaxent;
 
     if (strcmp(pszNewInstanceFName, pszInstanceFname) != 0) {
diff --git a/training/const_reorder/constituent_reorder_model.cc b/training/const_reorder/constituent_reorder_model.cc
index 6bec3f0b..d3ad0f2b 100644
--- a/training/const_reorder/constituent_reorder_model.cc
+++ b/training/const_reorder/constituent_reorder_model.cc
@@ -12,7 +12,7 @@
 
 #include "utils/filelib.h"
 
-#include "decoder/ff_const_reorder_common.h"
+#include "trainer.h"
 
 using namespace std;
 using namespace const_reorder;
@@ -104,8 +104,8 @@ struct SConstReorderTrainer {
 pZhangleMaxent->fnTrain(pszInstanceFname, "lbfgs", pszModelFname, 100, 2.0);
 delete pZhangleMaxent;*/
 
-    Tsuruoka_Maxent* pMaxent = new Tsuruoka_Maxent(NULL);
-    pMaxent->fnTrain(pszNewInstanceFName, "l1", pszModelFname, 300);
+    Tsuruoka_Maxent_Trainer* pMaxent = new Tsuruoka_Maxent_Trainer;
+    pMaxent->fnTrain(pszNewInstanceFName, "l1", pszModelFname);
     delete pMaxent;
 
     if (strcmp(pszNewInstanceFName, pszInstanceFname) != 0) {
diff --git a/training/const_reorder/trainer.cc b/training/const_reorder/trainer.cc
new file mode 100644
index 00000000..e22a8a66
--- /dev/null
+++ b/training/const_reorder/trainer.cc
@@ -0,0 +1,67 @@
+#include "trainer.h"
+
+Tsuruoka_Maxent_Trainer::Tsuruoka_Maxent_Trainer()
+    : const_reorder::Tsuruoka_Maxent(NULL) {}
+
+void Tsuruoka_Maxent_Trainer::fnTrain(const char* pszInstanceFName,
+                                      const char* pszAlgorithm,
+                                      const char* pszModelFName) {
+  assert(strcmp(pszAlgorithm, "l1") == 0 || strcmp(pszAlgorithm, "l2") == 0 ||
+         strcmp(pszAlgorithm, "sgd") == 0 || strcmp(pszAlgorithm, "SGD") == 0);
+  FILE* fpIn = fopen(pszInstanceFName, "r");
+
+  ME_Model* pModel = new ME_Model();
+
+  char* pszLine = new char[100001];
+  int iNumInstances = 0;
+  int iLen;
+  while (!feof(fpIn)) {
+    pszLine[0] = '\0';
+    fgets(pszLine, 20000, fpIn);
+    if (strlen(pszLine) == 0) {
+      continue;
+    }
+
+    iLen = strlen(pszLine);
+    while (iLen > 0 && pszLine[iLen - 1] > 0 && pszLine[iLen - 1] < 33) {
+      pszLine[iLen - 1] = '\0';
+      iLen--;
+    }
+
+    iNumInstances++;
+
+    ME_Sample* pmes = new ME_Sample();
+
+    char* p = strrchr(pszLine, ' ');
+    assert(p != NULL);
+    p[0] = '\0';
+    p++;
+    std::vector<std::string> vecContext;
+    SplitOnWhitespace(std::string(pszLine), &vecContext);
+
+    pmes->label = std::string(p);
+    for (size_t i = 0; i < vecContext.size(); i++)
+      pmes->add_feature(vecContext[i]);
+    pModel->add_training_sample((*pmes));
+    if (iNumInstances % 100000 == 0)
+      fprintf(stdout, "......Reading #Instances: %1d\n", iNumInstances);
+    delete pmes;
+  }
+  fprintf(stdout, "......Reading #Instances: %1d\n", iNumInstances);
+  fclose(fpIn);
+
+  if (strcmp(pszAlgorithm, "l1") == 0)
+    pModel->use_l1_regularizer(1.0);
+  else if (strcmp(pszAlgorithm, "l2") == 0)
+    pModel->use_l2_regularizer(1.0);
+  else
+    pModel->use_SGD();
+
+  pModel->train();
+  pModel->save_to_file(pszModelFName);
+
+  delete pModel;
+  fprintf(stdout, "......Finished Training\n");
+  fprintf(stdout, "......Model saved as %s\n", pszModelFName);
+  delete[] pszLine;
+}
diff --git a/training/const_reorder/trainer.h b/training/const_reorder/trainer.h
new file mode 100644
index 00000000..e574a536
--- /dev/null
+++ b/training/const_reorder/trainer.h
@@ -0,0 +1,12 @@
+#ifndef TRAINING_CONST_REORDER_TRAINER_H_
+#define TRAINING_CONST_REORDER_TRAINER_H_
+
+#include "decoder/ff_const_reorder_common.h"
+
+struct Tsuruoka_Maxent_Trainer : const_reorder::Tsuruoka_Maxent {
+  Tsuruoka_Maxent_Trainer();
+  void fnTrain(const char* pszInstanceFName, const char* pszAlgorithm,
+               const char* pszModelFName);
+};
+
+#endif  // TRAINING_CONST_REORDER_TRAINER_H_