Finished implementing the dynamical oracle

MLP/include/MLP.hpp

@@ -12,7 +12,6 @@
 #include <dynet/timing.h>
 #include <dynet/expr.h>
 #include "FeatureModel.hpp"
-#include "TrainingExamples.hpp"
 
 /// @brief Multi Layer Perceptron.
 ///

trainer/include/Trainer.hpp

@@ -9,7 +9,6 @@
 #include "TransitionMachine.hpp"
 #include "BD.hpp"
 #include "Config.hpp"
-#include "TrainingExamples.hpp"
 
 /// @brief An object capable of training a TransitionMachine given a BD initialized with training examples.
 class Trainer
@@ -68,17 +67,6 @@ class Trainer
 
   private :
 
-  /// @brief Train the TransitionMachine one example at a time.
-  ///
-  /// For each epoch all the Classifier of the TransitionMachine are fed all the 
-  /// training examples, at the end of the epoch Classifier are evaluated on 
-  /// the devBD if available, and each Classifier will be saved only if its score
-  /// on the current epoch is its all time best.\n
-  /// When a Classifier is saved that way, all the Dict involved are also saved.
-  /// @param nbIter The number of epochs.
-  /// @param mustShuffle Whether or not to shuffle examples between each epoch.
-  void trainUnbatched(int nbIter, bool mustShuffle);
-
   /// @brief Compute and print scores for each Classifier on this epoch, and save the Classifier if they achieved their all time best score.
   void printScoresAndSave(FILE * output);
 
@@ -107,13 +95,17 @@ class Trainer
   /// @param debugMode If true, infos will be printed on stderr.
   /// @param removeDuplicates If true, duplicates examples will be removed from the training set.
   Trainer(TransitionMachine & tm, BD & bd, Config & config, BD * devBD, Config * devConfig, bool debugMode, bool removeDuplicates);
-  /// @brief Train the TransitionMachine.
+
+  /// @brief Train the TransitionMachine one example at a time.
   ///
-  /// @param nbIter The number of training epochs.
-  /// @param batchSize The size of each batch.
-  /// @param mustShuffle Will the examples be shuffled after every epoch ?
-  /// @param batched True if we feed the training algorithm with batches of examples
-  void train(int nbIter, int batchSize, bool mustShuffle, bool batched);
+  /// For each epoch all the Classifier of the TransitionMachine are fed all the 
+  /// training examples, at the end of the epoch Classifier are evaluated on 
+  /// the devBD if available, and each Classifier will be saved only if its score
+  /// on the current epoch is its all time best.\n
+  /// When a Classifier is saved that way, all the Dict involved are also saved.
+  /// @param nbIter The number of epochs.
+  /// @param mustShuffle Whether or not to shuffle examples between each epoch.
+  void train(int nbIter, bool mustShuffle);
 };
 
 #endif

trainer/src/Trainer.cpp

@@ -26,6 +26,9 @@ std::map<std::string, float> Trainer::getScoreOnDev()
 
   std::map< std::string, std::pair<int, int> > counts;
 
+  if (debugMode)
+    fprintf(stderr, "Computing score on dev set\n");
+
   while (!devConfig->isFinal())
   {
     TransitionMachine::State * currentState = tm.getCurrentState();
@@ -73,6 +76,12 @@ std::map<std::string, float> Trainer::getScoreOnDev()
       std::string actionName = pAction;
       Action * action = classifier->getAction(actionName);
 
+      if (debugMode)
+      {
+        devConfig->printForDebug(stderr);
+        fprintf(stderr, "pAction=<%s> action=<%s>\n", pAction.c_str(), actionName.c_str());
+      }
+
       action->apply(*devConfig);
       TransitionMachine::Transition * transition = tm.getTransition(actionName);
       tm.takeTransition(transition);
@@ -87,7 +96,7 @@ std::map<std::string, float> Trainer::getScoreOnDev()
   return scores;
 }
 
-void Trainer::trainUnbatched(int nbIter, bool mustShuffle)
+void Trainer::train(int nbIter, bool mustShuffle)
 {
   this->nbIter = nbIter;
 
@@ -98,6 +107,7 @@ void Trainer::trainUnbatched(int nbIter, bool mustShuffle)
   for (curIter = 0; curIter < nbIter; curIter++)
   {
     tm.reset();
+
     trainConfig.reset();
 
     if(mustShuffle)
@@ -144,6 +154,7 @@ void Trainer::trainUnbatched(int nbIter, bool mustShuffle)
         if (zeroCostActions.empty())
         {
           fprintf(stderr, "ERROR (%s) : Unable to find any zero cost action. Aborting.\n", ERRINFO);
+          fprintf(stderr, "State : %s\n", currentState->name.c_str());
           trainConfig.printForDebug(stderr);
           exit(1);
         }
@@ -278,8 +289,3 @@ void Trainer::printScoresAndSave(FILE * output)
   printColumns(output, {names, acc, train, dev, savedStr});
 }
 
-void Trainer::train(int nbIter, int batchSize, bool mustShuffle, bool batched)
-{
-  trainUnbatched(nbIter, mustShuffle);
-}
-

trainer/src/macaon_train.cpp

@@ -43,14 +43,10 @@ po::options_description getOptionsDescription()
       "Language you are working with")
     ("nbiter,n", po::value<int>()->default_value(5),
       "Number of training epochs (iterations)")
-    ("batchsize,b", po::value<int>()->default_value(256),
-      "Size of each training batch (in number of examples)")
     ("seed,s", po::value<int>()->default_value(100),
       "The random seed that will initialize RNG")
     ("duplicates", po::value<bool>()->default_value(true),
       "Remove identical training examples")
-    ("batched", po::value<bool>()->default_value(true),
-      "Uses batch of training examples")
     ("shuffle", po::value<bool>()->default_value(true),
       "Shuffle examples after each iteration");
 
@@ -115,10 +111,8 @@ int main(int argc, char * argv[])
   std::string expName = vm["expName"].as<std::string>();
   std::string lang = vm["lang"].as<std::string>();
   int nbIter = vm["nbiter"].as<int>();
-  int batchSize = vm["batchsize"].as<int>();
   int randomSeed = vm["seed"].as<int>();
   bool mustShuffle = vm["shuffle"].as<bool>();
-  bool batched = vm["batched"].as<bool>();
   bool removeDuplicates = vm["duplicates"].as<bool>();
   bool debugMode = vm.count("debug") == 0 ? false : true;
 
@@ -159,7 +153,7 @@ int main(int argc, char * argv[])
   }
 
   trainer->expPath = expPath;
-  trainer->train(nbIter, batchSize, mustShuffle, batched);
+  trainer->train(nbIter, mustShuffle);
 
   return 0;
 }

transition_machine/include/TrainingExamples.hpp

-/// @file TrainingExamples.hpp
-/// @author Franck Dary
-/// @version 1.0
-/// @date 2018-08-09
-
-#ifndef TRAININGEXAMPLES__H
-#define TRAININGEXAMPLES__H
-
-#include <vector>
-#include "FeatureModel.hpp"
-
-class TrainingExamples
-{
-  public :
-
-  std::vector<unsigned int> order;
-  std::vector<FeatureModel::FeatureDescription> examples;
-  std::vector<int> classes;
-  int nextIndex;
-
-  public :
-
-  TrainingExamples();
-  void add(const FeatureModel::FeatureDescription & example, int gold);
-  unsigned int size();
-  TrainingExamples getBatch(unsigned int batchSize);
-  void reset();
-  void shuffle();
-  void remove(int index);
-  void removeDuplicates();
-};
-
-#endif

transition_machine/src/ActionBank.cpp

@@ -102,8 +102,8 @@ std::vector<Action::BasicAction> ActionBank::str2sequence(const std::string & na
       {c.stack.push_back(c.head);};
     auto undo = [](Config & c, Action::BasicAction &)
       {c.stack.pop_back();};
-    auto appliable = [](Config &, Action::BasicAction &)
-      {return true;};
+    auto appliable = [](Config & c, Action::BasicAction &)
+      {return c.head < (int)c.tapes[0].ref.size()-1;};
     Action::BasicAction basicAction =
       {Action::BasicAction::Type::Push, "", apply, undo, appliable};
 

transition_machine/src/Config.cpp

@@ -56,6 +56,9 @@ void Config::readInput(const std::string & filename)
       tape.ref.emplace_back();
 
     tape.hyp.resize(tape.ref.size());
+
+    tape.ref.emplace_back("0");
+    tape.hyp.emplace_back("");
   }
 }
 
@@ -134,9 +137,9 @@ void Config::printAsOutput(FILE * output)
     if(bd.mustPrintLine(j))
       lastToPrint = j;
 
-  for (unsigned int i = 0; i < tapes[0].hyp.size(); i++)
-  {
+  for (unsigned int i = 0; i < tapes[0].hyp.size() - 1; i++)
     for (unsigned int j = 0; j < tapes.size(); j++)
+    {
       if(bd.mustPrintLine(j))
         fprintf(output, "%s%s", tapes[j][i].empty() ? "0" : tapes[j][i].c_str(), j == lastToPrint ? "\n" : "\t");
     }
@@ -144,13 +147,13 @@ void Config::printAsOutput(FILE * output)
 
 void Config::moveHead(int mvt)
 {
-  if (mvt + head < (int)getTapeByInputCol(0).hyp.size())
+  if (head + mvt < (int)tapes[0].ref.size())
     head += mvt;
 }
 
 bool Config::isFinal()
 {
-  return head >= (int)getTapeByInputCol(0).hyp.size() -1;
+  return head >= (int)getTapeByInputCol(0).hyp.size()-1 && stack.empty();
 }
 
 void Config::reset()
@@ -210,6 +213,8 @@ void Config::shuffle(const std::string & delimiterTape, const std::string & deli
       previousIndex = i+1;
     }
 
+  std::pair<unsigned int, unsigned int> suffix = {delimiters.back().second+1, tape.ref.size()-1};
+
   std::random_shuffle(delimiters.begin(), delimiters.end());
 
   std::vector<Tape> newTapes = tapes;
@@ -224,6 +229,12 @@ void Config::shuffle(const std::string & delimiterTape, const std::string & deli
       std::copy(tapes[tape].ref.begin()+delimiter.first, tapes[tape].ref.begin()+delimiter.second+1, std::back_inserter(newTapes[tape].ref));
       std::copy(tapes[tape].hyp.begin()+delimiter.first, tapes[tape].hyp.begin()+delimiter.second+1, std::back_inserter(newTapes[tape].hyp));
     }
+
+    if (suffix.first <= suffix.second)
+    {
+      std::copy(tapes[tape].ref.begin()+suffix.first, tapes[tape].ref.begin()+suffix.second+1, std::back_inserter(newTapes[tape].ref));
+      std::copy(tapes[tape].hyp.begin()+suffix.first, tapes[tape].hyp.begin()+suffix.second+1, std::back_inserter(newTapes[tape].hyp));
+    }
   }
 
   tapes = newTapes;

transition_machine/src/Oracle.cpp

@@ -86,7 +86,7 @@ void Oracle::createDatabase()
   },
   [](Config & c, Oracle *, const std::string & action)
   {
-    return action == "WRITE 0 POS " + c.getTape("POS").ref[c.head];
+    return action == "WRITE 0 POS " + c.getTape("POS").ref[c.head] || c.head >= (int)c.tapes[0].ref.size()-1;
   })));
 
   str2oracle.emplace("morpho", std::unique_ptr<Oracle>(new Oracle(
@@ -102,7 +102,7 @@ void Oracle::createDatabase()
   },
   [](Config & c, Oracle *, const std::string & action)
   {
-    return action == "WRITE 0 MORPHO " + c.getTape("MORPHO").ref[c.head];
+    return action == "WRITE 0 MORPHO " + c.getTape("MORPHO").ref[c.head] || c.head >= (int)c.tapes[0].ref.size()-1;
   })));
 
   str2oracle.emplace("signature", std::unique_ptr<Oracle>(new Oracle(
@@ -211,7 +211,7 @@ void Oracle::createDatabase()
     const std::string & lemma = c.getTape("LEMMA").ref[c.head];
     std::string rule = getRule(form, lemma);
 
-    return action == std::string("RULE LEMMA ON FORM ") + rule;
+    return action == std::string("RULE LEMMA ON FORM ") + rule || c.head >= (int)c.tapes[0].ref.size()-1;
   })));
 
   str2oracle.emplace("parser", std::unique_ptr<Oracle>(new Oracle(
@@ -244,6 +244,8 @@ void Oracle::createDatabase()
       sentenceStart++;
     while(sentenceEnd < (int)eos.ref.size() && eos.ref[sentenceEnd] != "1")
       sentenceEnd++;
+    if (sentenceEnd == (int)eos.ref.size())
+      sentenceEnd--;
 
     auto parts = split(action);
 

transition_machine/src/TrainingExamples.cpp

-#include "TrainingExamples.hpp" 
-#include <algorithm>
-
-TrainingExamples::TrainingExamples()
-{
-  nextIndex = 0;
-}
-
-void TrainingExamples::add(const FeatureModel::FeatureDescription & example, int gold)
-{
-  examples.emplace_back(example);
-  classes.emplace_back(gold);
-  order.emplace_back(order.size());
-}
-
-unsigned int TrainingExamples::size()
-{
-  return order.size();
-}
-
-TrainingExamples TrainingExamples::getBatch(unsigned int batchSize)
-{
-  TrainingExamples batch;
-
-  for(unsigned int i = 0; i < batchSize && (unsigned)nextIndex < order.size()-1; i++)
-  {
-    batch.add(examples[order[nextIndex]], classes[order[nextIndex]]);
-    nextIndex++;
-  }
-
-  return batch;
-}
-
-void TrainingExamples::reset()
-{
-  nextIndex = 0;
-}
-
-void TrainingExamples::shuffle()
-{
-  std::random_shuffle(order.begin(), order.end());
-}
-
-void TrainingExamples::removeDuplicates()
-{
-  std::map<std::string, int> lastIndex;
-  std::map<int, bool> toRemove;
-
-  for (unsigned int i = 0; i < examples.size(); i++)
-  {
-    std::string example = examples[i].toString();
-
-    if (lastIndex.count(example))
-      toRemove[i] = true;
-    else
-      lastIndex[example] = i;
-  }
-
-  for (auto & it : toRemove)
-    remove(it.first);
-}
-
-void TrainingExamples::remove(int index)
-{
-  for (unsigned int i = 0; i < order.size(); i++)
-    if ((int)order[i] == index)
-    {
-      order[i] = order.back();
-      order.pop_back();
-    }
-}
-