diff --git a/decoder/include/Decoder.hpp b/decoder/include/Decoder.hpp
index bdec405e338d3af97b03b14bdd960ca493b3d114..01bc7cc5b70e7caaef1d6d556c9ace4d4c1bc320 100644
--- a/decoder/include/Decoder.hpp
+++ b/decoder/include/Decoder.hpp
@@ -26,7 +26,7 @@ class Decoder
Decoder(ReadingMachine & machine);
void decode(BaseConfig & config, std::size_t beamSize, float beamThreshold, bool debug, bool printAdvancement);
- void evaluate(const Config & config, std::filesystem::path modelPath, const std::string goldTSV, const std::set<std::string> & predicted);
+ void evaluate(const std::vector<const Config *> & configs, std::filesystem::path modelPath, const std::string goldTSV, const std::set<std::string> & predicted);
std::vector<std::pair<float,std::string>> getF1Scores(const std::set<std::string> & colNames) const;
std::vector<std::pair<float,std::string>> getAlignedAccs(const std::set<std::string> & colNames) const;
std::vector<std::pair<float,std::string>> getRecalls(const std::set<std::string> & colNames) const;
diff --git a/decoder/src/Decoder.cpp b/decoder/src/Decoder.cpp
index 73d8984244ef05f069c989feaaaae0975888939a..474db0788b45b047b03b16905bfab0f9ed9f8dce 100644
--- a/decoder/src/Decoder.cpp
+++ b/decoder/src/Decoder.cpp
@@ -135,12 +135,13 @@ std::string Decoder::getMetricOfColName(const std::string & colName) const
return colName;
}
-void Decoder::evaluate(const Config & config, std::filesystem::path modelPath, const std::string goldTSV, const std::set<std::string> & predicted)
+void Decoder::evaluate(const std::vector<const Config *> & configs, std::filesystem::path modelPath, const std::string goldTSV, const std::set<std::string> & predicted)
{
evaluation.clear();
auto predictedTSV = (modelPath/"predicted_dev.tsv").string();
std::FILE * predictedTSVFile = std::fopen(predictedTSV.c_str(), "w");
- config.print(predictedTSVFile);
+ for (unsigned int i = 0; i < configs.size(); i++)
+ configs[i]->print(predictedTSVFile, i==0);
std::fclose(predictedTSVFile);
std::FILE * evalFromUD = popen(fmt::format("{} {} {} -x {}", "../scripts/conll18_ud_eval.py", goldTSV, predictedTSV, util::join(",", std::vector<std::string>(predicted.begin(), predicted.end()))).c_str(), "r");
diff --git a/decoder/src/MacaonDecode.cpp b/decoder/src/MacaonDecode.cpp
index 0bbf5aab3a6a41bf0835581accf861e409341d29..e4194e84005028034e55ccb79e11cd9d10d3b1cc 100644
--- a/decoder/src/MacaonDecode.cpp
+++ b/decoder/src/MacaonDecode.cpp
@@ -98,16 +98,19 @@ int MacaonDecode::main()
if (inputTXT.empty())
{
- BaseConfig config(mcd, inputTSV, util::utf8string());
+ BaseConfig config(mcd, inputTSV, util::utf8string(), std::vector<int>());
decoder.decode(config, beamSize, beamThreshold, debug, printAdvancement);
config.print(stdout, true);
}
else
{
auto inputs = util::readFileAsUtf8(inputTXT, lineByLine);
- for (unsigned int i = 0; i < inputs.size(); i++)
+ for (int i = 0; i < (int)inputs.size(); i++)
{
- BaseConfig config(mcd, inputTSV, inputs[i]);
+ std::vector<int> sentIndexes;
+ if (inputs.size() > 1)
+ sentIndexes.emplace_back(i);
+ BaseConfig config(mcd, inputTSV, inputs[i], sentIndexes);
decoder.decode(config, beamSize, beamThreshold, debug, printAdvancement);
config.print(stdout, i == 0);
}
diff --git a/reading_machine/include/BaseConfig.hpp b/reading_machine/include/BaseConfig.hpp
index d77d0a0720ed14439412258a6fa9813514aac173..a93b1804dfe23a470dd1794a8338a390a923a408 100644
--- a/reading_machine/include/BaseConfig.hpp
+++ b/reading_machine/include/BaseConfig.hpp
@@ -21,11 +21,11 @@ class BaseConfig : public Config
void createColumns(std::string mcd);
void readRawInput(std::string_view rawFilename);
- void readTSVInput(std::string_view tsvFilename);
+ void readTSVInput(std::string_view tsvFilename, const std::vector<int> & sentencesIndexes);
public :
- BaseConfig(std::string mcd, std::string_view tsvFilename, const util::utf8string & rawFilename);
+ BaseConfig(std::string mcd, std::string_view tsvFilename, const util::utf8string & rawFilename, const std::vector<int> & sentencesIndexes);
BaseConfig(const BaseConfig & other);
BaseConfig & operator=(const BaseConfig & other) = default;
diff --git a/reading_machine/src/BaseConfig.cpp b/reading_machine/src/BaseConfig.cpp
index 298d4537b064b4126d7f67e04129abceeaf6529c..c9956d4886037ce308840293f6d800c0b7a3272d 100644
--- a/reading_machine/src/BaseConfig.cpp
+++ b/reading_machine/src/BaseConfig.cpp
@@ -43,125 +43,142 @@ void BaseConfig::readRawInput(std::string_view rawFilename)
rawInput.replace(util::utf8char("\t"), util::utf8char(" "));
}
-void BaseConfig::readTSVInput(std::string_view tsvFilename)
+void BaseConfig::readTSVInput(std::string_view tsvFilename, const std::vector<int> & _sentencesIndexes)
{
- std::FILE * file = std::fopen(tsvFilename.data(), "r");
+ auto sentencesIndexes = _sentencesIndexes;
+ if (sentencesIndexes.empty())
+ sentencesIndexes.emplace_back(-1);
- if (not file)
- util::myThrow(fmt::format("Cannot open file '{}'", tsvFilename));
-
- char lineBuffer[100000];
- int inputLineIndex = 0;
- bool inputHasBeenRead = false;
- int usualNbCol = -1;
- int nbMultiwords = 0;
- std::vector<std::string> pendingComments;
-
- while (!std::feof(file))
+ for (int targetSentenceIndex : sentencesIndexes)
{
- if (lineBuffer != std::fgets(lineBuffer, 100000, file))
- break;
-
- std::string_view line(lineBuffer);
- inputLineIndex++;
-
- if (line.size() < 3)
+ std::FILE * file = std::fopen(tsvFilename.data(), "r");
+
+ if (not file)
+ util::myThrow(fmt::format("Cannot open file '{}'", tsvFilename));
+
+ char lineBuffer[100000];
+ int inputLineIndex = 0;
+ bool inputHasBeenRead = false;
+ int usualNbCol = -1;
+ int nbMultiwords = 0;
+ int curSentenceIndex = 0;
+ std::vector<std::string> pendingComments;
+
+ while (!std::feof(file))
{
- if (!inputHasBeenRead)
- continue;
-
- get(EOSColName, getNbLines()-1, 0) = EOSSymbol1;
-
- try
+ if (lineBuffer != std::fgets(lineBuffer, 100000, file))
+ break;
+
+ std::string_view line(lineBuffer);
+ inputLineIndex++;
+
+ if (line.size() < 3)
{
- std::map<std::string, int> id2index;
- int firstIndexOfSequence = getNbLines()-1;
- for (int i = (int)getNbLines()-1; has(0, i, 0); --i)
- {
- if (!isToken(i))
- continue;
+ if (!inputHasBeenRead)
+ continue;
- if (i != (int)getNbLines()-1 && getConst(EOSColName, i, 0) == EOSSymbol1)
- break;
-
- firstIndexOfSequence = i;
- id2index[getConst(idColName, i, 0)] = i;
- }
- if (hasColIndex(headColName))
- for (int i = firstIndexOfSequence; i < (int)getNbLines(); ++i)
+ if (targetSentenceIndex == -1 or targetSentenceIndex == curSentenceIndex)
+ {
+ get(EOSColName, getNbLines()-1, 0) = EOSSymbol1;
+
+ try
{
- if (!isToken(i))
- continue;
- auto & head = get(headColName, i, 0);
- if (head == "0")
- head = "-1";
- else
- head = std::to_string(id2index[head]);
- }
- } catch(std::exception & e) {util::myThrow(e.what());}
-
- continue;
- }
-
- if (line.back() == '\n')
- line.remove_suffix(1);
-
- if (line[0] == '#')
- {
- if (util::doIfNameMatch(std::regex("(?:(?:\\s|\\t)*)#(?:(?:\\s|\\t)*)global.columns(?:(?:\\s|\\t)*)=(?:(?:\\s|\\t)*)(.+)"), line, [this](const auto & sm)
+ std::map<std::string, int> id2index;
+ int firstIndexOfSequence = getNbLines()-1;
+ for (int i = (int)getNbLines()-1; has(0, i, 0); --i)
{
- createColumns(util::join(",", util::split(util::strip(sm.str(1)), ' ')));
- }))
- continue;
-
- pendingComments.emplace_back(line);
- continue;
- }
-
- inputHasBeenRead = true;
-
- auto splited = util::split(line, '\t');
- if (usualNbCol == -1)
- usualNbCol = splited.size();
- if ((int)splited.size() != usualNbCol)
- util::myThrow(fmt::format("in file {} line {} is invalid, it shoud have {} columns", tsvFilename, line, usualNbCol));
-
- // Ignore empty nodes
- if (hasColIndex(idColName) && splited[getColIndex(idColName)].find('.') != std::string::npos)
- continue;
-
- addLines(1);
- get(EOSColName, getNbLines()-1, 0) = EOSSymbol0;
- if (nbMultiwords > 0)
- {
- get(isMultiColName, getNbLines()-1, 0) = EOSSymbol1;
- nbMultiwords--;
- }
- else
- get(isMultiColName, getNbLines()-1, 0) = EOSSymbol0;
-
- get(commentsColName, getNbLines()-1, 0) = util::join("\n", pendingComments);
- pendingComments.clear();
+ if (!isToken(i))
+ continue;
+
+ if (i != (int)getNbLines()-1 && getConst(EOSColName, i, 0) == EOSSymbol1)
+ break;
+
+ firstIndexOfSequence = i;
+ id2index[getConst(idColName, i, 0)] = i;
+ }
+ if (hasColIndex(headColName))
+ for (int i = firstIndexOfSequence; i < (int)getNbLines(); ++i)
+ {
+ if (!isToken(i))
+ continue;
+ auto & head = get(headColName, i, 0);
+ if (head == "0")
+ head = "-1";
+ else
+ head = std::to_string(id2index[head]);
+ }
+ } catch(std::exception & e) {util::myThrow(e.what());}
+ }
- for (unsigned int i = 0; i < splited.size(); i++)
- if (i < colIndex2Name.size() - extraColumns.size())
+ curSentenceIndex += 1;
+
+ continue;
+ }
+
+ if (line.back() == '\n')
+ line.remove_suffix(1);
+
+ if (line[0] == '#')
{
- std::string value = std::string(splited[i]);
- get(i, getNbLines()-1, 0) = value;
+ if (util::doIfNameMatch(std::regex("(?:(?:\\s|\\t)*)#(?:(?:\\s|\\t)*)global.columns(?:(?:\\s|\\t)*)=(?:(?:\\s|\\t)*)(.+)"), line, [this](const auto & sm)
+ {
+ createColumns(util::join(",", util::split(util::strip(sm.str(1)), ' ')));
+ }))
+ continue;
+
+ if (targetSentenceIndex == -1 or targetSentenceIndex == curSentenceIndex)
+ pendingComments.emplace_back(line);
+ continue;
}
+
+ inputHasBeenRead = true;
+
+ auto splited = util::split(line, '\t');
+ if (usualNbCol == -1)
+ usualNbCol = splited.size();
+ if ((int)splited.size() != usualNbCol)
+ util::myThrow(fmt::format("in file {} line {} is invalid, it shoud have {} columns", tsvFilename, line, usualNbCol));
+
+ if (targetSentenceIndex != -1 and targetSentenceIndex != curSentenceIndex)
+ continue;
- if (isMultiword(getNbLines()-1))
- nbMultiwords = getMultiwordSize(getNbLines()-1)+1;
- }
-
- std::fclose(file);
+ // Ignore empty nodes
+ if (hasColIndex(idColName) && splited[getColIndex(idColName)].find('.') != std::string::npos)
+ continue;
+
+ addLines(1);
+ get(EOSColName, getNbLines()-1, 0) = EOSSymbol0;
+ if (nbMultiwords > 0)
+ {
+ get(isMultiColName, getNbLines()-1, 0) = EOSSymbol1;
+ nbMultiwords--;
+ }
+ else
+ get(isMultiColName, getNbLines()-1, 0) = EOSSymbol0;
+
+ get(commentsColName, getNbLines()-1, 0) = util::join("\n", pendingComments);
+ pendingComments.clear();
+
+ for (unsigned int i = 0; i < splited.size(); i++)
+ if (i < colIndex2Name.size() - extraColumns.size())
+ {
+ std::string value = std::string(splited[i]);
+ get(i, getNbLines()-1, 0) = value;
+ }
+
+ if (isMultiword(getNbLines()-1))
+ nbMultiwords = getMultiwordSize(getNbLines()-1)+1;
+ }
+
+ std::fclose(file);
+ } // End for targetSentenceIndex
}
BaseConfig::BaseConfig(const BaseConfig & other) : Config(other), colIndex2Name(other.colIndex2Name), colName2Index(other.colName2Index)
{
}
-BaseConfig::BaseConfig(std::string mcd, std::string_view tsvFilename, const util::utf8string & rawInput)
+BaseConfig::BaseConfig(std::string mcd, std::string_view tsvFilename, const util::utf8string & rawInput, const std::vector<int> & sentencesIndexes)
{
if (tsvFilename.empty() and rawInput.empty())
util::myThrow("tsvFilename and rawInput can't be both empty");
@@ -171,8 +188,9 @@ BaseConfig::BaseConfig(std::string mcd, std::string_view tsvFilename, const util
if (not rawInput.empty())
this->rawInput = rawInput;
+// sentencesIndexes = index of sentences to keep. Empty vector == keep all sentences.
if (not tsvFilename.empty())
- readTSVInput(tsvFilename);
+ readTSVInput(tsvFilename, sentencesIndexes);
if (!has(0,wordIndex,0))
addLines(1);
diff --git a/reading_machine/src/Config.cpp b/reading_machine/src/Config.cpp
index ddc5477fb58fce6e3e465e5549ea7f31ea285bbc..3268d77c1564bf6d1d90af4d35475ed060525410 100644
--- a/reading_machine/src/Config.cpp
+++ b/reading_machine/src/Config.cpp
@@ -11,7 +11,7 @@ Config::Config(const Config & other)
this->appliableSplitTransitions = other.appliableSplitTransitions;
this->appliableTransitions = other.appliableTransitions;
- this->strategy.reset(new Strategy(*other.strategy));
+ this->strategy.reset(other.strategy ? new Strategy(*other.strategy) : nullptr);
this->rawInput = other.rawInput;
this->wordIndex = other.wordIndex;
@@ -45,6 +45,9 @@ void Config::resizeLines(unsigned int nbLines)
bool Config::has(int colIndex, int lineIndex, int hypothesisIndex) const
{
+// fmt::print(stderr, "line index = {}\n", lineIndex);
+// fmt::print(stderr, "first line index = {}\n", getFirstLineIndex());
+// fmt::print(stderr, "nbLines = {}\n", getNbLines());
return colIndex >= 0 && colIndex < (int)getNbColumns() && lineIndex >= (int)getFirstLineIndex() && lineIndex < (int)getFirstLineIndex() + (int)getNbLines() && hypothesisIndex >= 0 && hypothesisIndex < nbHypothesesMax+1;
}
@@ -230,7 +233,7 @@ void Config::printForDebug(FILE * dest) const
fmt::print(dest, "{}", getLetter(index));
if (!util::isEmpty(rawInput))
fmt::print(dest, "\n{}\n", longLine);
- fmt::print(dest, "State={}\nwordIndex={} characterIndex={}\nhistory=({})\nstack=({})\n", state, wordIndex, characterIndex, historyStr, stackStr);
+ fmt::print(dest, "State={}\nwordIndex={}/{} characterIndex={}/{}\nhistory=({})\nstack=({})\n", state, wordIndex, getNbLines(), characterIndex, rawInput.size(), historyStr, stackStr);
fmt::print(dest, "{}\n", longLine);
for (unsigned int line = 0; line < toPrint.size(); line++)
diff --git a/reading_machine/src/SubConfig.cpp b/reading_machine/src/SubConfig.cpp
index 7b7b65ead097e4bb1d0d33b0a5bbd9d8b55831a9..d9c191227e536239849ce348029ecff0ec2210dc 100644
--- a/reading_machine/src/SubConfig.cpp
+++ b/reading_machine/src/SubConfig.cpp
@@ -30,7 +30,7 @@ bool SubConfig::update()
{
unsigned int currentLastLineIndex = firstLineIndex + getNbLines();
- if (currentLastLineIndex >= model.getNbLines()-1)
+ if (getNbLines() > 0 and currentLastLineIndex >= model.getNbLines()-1)
return false;
std::size_t newFirstLineIndex = spanSize/2 >= wordIndex ? 0 : wordIndex - spanSize/2;
diff --git a/reading_machine/src/Transition.cpp b/reading_machine/src/Transition.cpp
index e80ae0b4252f5b935aaebf4e27d12766a576de80..e15546a8642baafaec0507283f41736be23d5e60 100644
--- a/reading_machine/src/Transition.cpp
+++ b/reading_machine/src/Transition.cpp
@@ -313,7 +313,7 @@ void Transition::initSplitWord(std::vector<std::string> words)
auto consumedWord = util::splitAsUtf8(words[0]);
sequence.emplace_back(Action::assertIsEmpty(Config::idColName, Config::Object::Buffer, 0));
sequence.emplace_back(Action::assertIsEmpty("FORM", Config::Object::Buffer, 0));
- sequence.emplace_back(Action::addLinesIfNeeded(words.size()));
+ sequence.emplace_back(Action::addLinesIfNeeded(words.size()-1));
sequence.emplace_back(Action::addCharsToCol("FORM", consumedWord.size(), Config::Object::Buffer, 0));
sequence.emplace_back(Action::consumeCharacterIndex(consumedWord));
for (unsigned int i = 1; i < words.size(); i++)
diff --git a/trainer/include/Trainer.hpp b/trainer/include/Trainer.hpp
index ef0f25ec18ab3be0ca9994a4f77dc30ec244adb7..dfa465e8a524134b55c7887892940fe0bc1a01cd 100644
--- a/trainer/include/Trainer.hpp
+++ b/trainer/include/Trainer.hpp
@@ -55,13 +55,13 @@ class Trainer
private :
- void extractExamples(SubConfig & config, bool debug, std::filesystem::path dir, int epoch, bool dynamicOracle, float explorationThreshold);
+ void extractExamples(std::vector<SubConfig> & configs, bool debug, std::filesystem::path dir, int epoch, bool dynamicOracle, float explorationThreshold);
float processDataset(DataLoader & loader, bool train, bool printAdvancement, int nbExamples);
public :
Trainer(ReadingMachine & machine, int batchSize);
- void createDataset(BaseConfig & goldConfig, bool debug, std::filesystem::path dir, int epoch, bool dynamicOracle, float explorationThreshold);
+ void createDataset(std::vector<BaseConfig> & goldConfigs, bool debug, std::filesystem::path dir, int epoch, bool dynamicOracle, float explorationThreshold);
void extractActionSequence(BaseConfig & config);
void makeDataLoader(std::filesystem::path dir);
void makeDevDataLoader(std::filesystem::path dir);
diff --git a/trainer/src/MacaonTrain.cpp b/trainer/src/MacaonTrain.cpp
index a0492974a6370f4f48a1d105664e1bcad129154b..dba21cd7549380cc0f5213f75c71c80f95b329fa 100644
--- a/trainer/src/MacaonTrain.cpp
+++ b/trainer/src/MacaonTrain.cpp
@@ -46,6 +46,7 @@ po::options_description MacaonTrain::getOptionsDescription()
("maxNorm", po::value<float>()->default_value(std::numeric_limits<float>::max()),
"Max norm for the embeddings")
("lockPretrained", "Disable fine tuning of all pretrained word embeddings.")
+ ("lineByLine", "Treat the TXT input as being one different text per line.")
("help,h", "Produce this help message")
("oracleMode", "Don't train a model, transform the corpus into a sequence of transitions.");
@@ -138,6 +139,7 @@ int MacaonTrain::main()
auto explorationThreshold = variables["explorationThreshold"].as<float>();
auto seed = variables["seed"].as<int>();
auto oracleMode = variables.count("oracleMode") == 0 ? false : true;
+ auto lineByLine = variables.count("lineByLine") == 0 ? false : true;
WordEmbeddingsImpl::setMaxNorm(variables["maxNorm"].as<float>());
WordEmbeddingsImpl::setScaleGradByFreq(variables.count("scaleGrad") != 0);
WordEmbeddingsImpl::setCanTrainPretrained(variables.count("lockPretrained") == 0);
@@ -165,27 +167,37 @@ int MacaonTrain::main()
ReadingMachine machine(machinePath.string(), true);
- util::utf8string trainRawInput;
+ std::vector<util::utf8string> trainRawInputs;
if (!trainRawFile.empty())
- {
- auto input = util::readFileAsUtf8(trainRawFile, false);
- trainRawInput = input[0];
- }
- BaseConfig goldConfig(mcd, trainTsvFile, trainRawInput);
- util::utf8string devRawInput;
- if (!devRawFile.empty())
- {
- auto input = util::readFileAsUtf8(devRawFile, false);
- devRawInput = input[0];
- }
- BaseConfig devGoldConfig(mcd, computeDevScore ? (devRawFile.empty() ? devTsvFile : "") : devTsvFile, devRawInput);
+ trainRawInputs = util::readFileAsUtf8(trainRawFile, lineByLine);
+ std::vector<BaseConfig> goldConfigs;
+ for (int i = 0; i < (int)trainRawInputs.size(); i++)
+ if (trainRawInputs.size() == 1)
+ goldConfigs.emplace_back(mcd, trainTsvFile, trainRawInputs[i], std::vector<int>());
+ else
+ goldConfigs.emplace_back(mcd, trainTsvFile, trainRawInputs[i], std::vector<int>{i});
+ if (goldConfigs.empty())
+ goldConfigs.emplace_back(mcd, trainTsvFile, util::utf8string(), std::vector<int>());
+ std::vector<util::utf8string> devRawInputs;
+ if (!devRawFile.empty())
+ devRawInputs = util::readFileAsUtf8(devRawFile, lineByLine);
+ std::vector<BaseConfig> devGoldConfigs;
+ for (int i = 0; i < (int)devRawInputs.size(); i++)
+ if (devRawInputs.size() == 1)
+ devGoldConfigs.emplace_back(mcd, computeDevScore ? (devRawFile.empty() ? devTsvFile : "") : devTsvFile, devRawInputs[i], std::vector<int>());
+ else
+ devGoldConfigs.emplace_back(mcd, computeDevScore ? (devRawFile.empty() ? devTsvFile : "") : devTsvFile, devRawInputs[i], std::vector<int>{i});
+ if (devGoldConfigs.empty())
+ devGoldConfigs.emplace_back(mcd, computeDevScore ? (devRawFile.empty() ? devTsvFile : "") : devTsvFile, util::utf8string(), std::vector<int>());
+
Trainer trainer(machine, batchSize);
Decoder decoder(machine);
if (oracleMode)
{
- trainer.extractActionSequence(goldConfig);
+ //TODO : handle more than one
+ trainer.extractActionSequence(goldConfigs[0]);
exit(0);
}
@@ -230,11 +242,11 @@ int MacaonTrain::main()
if (trainStrategy[currentEpoch].count(Trainer::TrainAction::ExtractGold) or trainStrategy[currentEpoch].count(Trainer::TrainAction::ExtractDynamic))
{
machine.setDictsState(trainStrategy[currentEpoch].count(Trainer::TrainAction::ExtractDynamic) ? Dict::State::Closed : Dict::State::Open);
- trainer.createDataset(goldConfig, debug, modelPath/"examples/train", currentEpoch, trainStrategy[currentEpoch].count(Trainer::TrainAction::ExtractDynamic), explorationThreshold);
+ trainer.createDataset(goldConfigs, debug, modelPath/"examples/train", currentEpoch, trainStrategy[currentEpoch].count(Trainer::TrainAction::ExtractDynamic), explorationThreshold);
if (!computeDevScore)
{
machine.setDictsState(Dict::State::Closed);
- trainer.createDataset(devGoldConfig, debug, modelPath/"examples/dev", currentEpoch, trainStrategy[currentEpoch].count(Trainer::TrainAction::ExtractDynamic), explorationThreshold);
+ trainer.createDataset(devGoldConfigs, debug, modelPath/"examples/dev", currentEpoch, trainStrategy[currentEpoch].count(Trainer::TrainAction::ExtractDynamic), explorationThreshold);
}
}
if (trainStrategy[currentEpoch].count(Trainer::TrainAction::ResetParameters) or trainStrategy[currentEpoch].count(Trainer::TrainAction::ResetOptimizer))
@@ -263,9 +275,21 @@ int MacaonTrain::main()
std::vector<std::pair<float,std::string>> devScores;
if (computeDevScore)
{
- BaseConfig devConfig(mcd, computeDevScore ? (devRawFile.empty() ? devTsvFile : "") : devTsvFile, devRawInput);
- decoder.decode(devConfig, 1, 0.0, debug, printAdvancement);
- decoder.evaluate(devConfig, modelPath, devTsvFile, machine.getPredicted());
+ std::vector<BaseConfig> devConfigs;
+ for (int i = 0; i < (int)devRawInputs.size(); i++)
+ if (devRawInputs.size() == 1)
+ devConfigs.emplace_back(mcd, computeDevScore ? (devRawFile.empty() ? devTsvFile : "") : devTsvFile, devRawInputs[i], std::vector<int>());
+ else
+ devConfigs.emplace_back(mcd, computeDevScore ? (devRawFile.empty() ? devTsvFile : "") : devTsvFile, devRawInputs[i], std::vector<int>{i});
+ if (devConfigs.empty())
+ devConfigs.emplace_back(mcd, computeDevScore ? (devRawFile.empty() ? devTsvFile : "") : devTsvFile, util::utf8string(), std::vector<int>());
+ for (auto & devConfig : devConfigs)
+ decoder.decode(devConfig, 1, 0.0, debug, printAdvancement);
+
+ std::vector<const Config *> devConfigsPtrs;
+ for (auto & devConfig : devConfigs)
+ devConfigsPtrs.emplace_back(&devConfig);
+ decoder.evaluate(devConfigsPtrs, modelPath, devTsvFile, machine.getPredicted());
devScores = decoder.getF1Scores(machine.getPredicted());
}
else
diff --git a/trainer/src/Trainer.cpp b/trainer/src/Trainer.cpp
index 5b30270013a9b1807c356fc1a6399ef1db9a6cd2..929d8620f530127c313378173d0c423de6e44151 100644
--- a/trainer/src/Trainer.cpp
+++ b/trainer/src/Trainer.cpp
@@ -17,18 +17,20 @@ void Trainer::makeDevDataLoader(std::filesystem::path dir)
devDataLoader = torch::data::make_data_loader(*devDataset, torch::data::DataLoaderOptions(batchSize).workers(0).max_jobs(0));
}
-void Trainer::createDataset(BaseConfig & goldConfig, bool debug, std::filesystem::path dir, int epoch, bool dynamicOracle, float explorationThreshold)
+void Trainer::createDataset(std::vector<BaseConfig> & goldConfigs, bool debug, std::filesystem::path dir, int epoch, bool dynamicOracle, float explorationThreshold)
{
- SubConfig config(goldConfig, goldConfig.getNbLines());
+ std::vector<SubConfig> configs;
+ for (auto & goldConfig : goldConfigs)
+ configs.emplace_back(goldConfig, goldConfig.getNbLines());
machine.trainMode(false);
- extractExamples(config, debug, dir, epoch, dynamicOracle, explorationThreshold);
+ extractExamples(configs, debug, dir, epoch, dynamicOracle, explorationThreshold);
machine.saveDicts();
}
-void Trainer::extractExamples(SubConfig & config, bool debug, std::filesystem::path dir, int epoch, bool dynamicOracle, float explorationThreshold)
+void Trainer::extractExamples(std::vector<SubConfig> & configs, bool debug, std::filesystem::path dir, int epoch, bool dynamicOracle, float explorationThreshold)
{
torch::AutoGradMode useGrad(false);
@@ -37,11 +39,6 @@ void Trainer::extractExamples(SubConfig & config, bool debug, std::filesystem::p
std::filesystem::create_directories(dir);
- config.addPredicted(machine.getPredicted());
- config.setStrategy(machine.getStrategyDefinition());
- config.setState(config.getStrategy().getInitialState());
- machine.getClassifier(config.getState())->setState(config.getState());
-
auto currentEpochAllExtractedFile = dir / fmt::format("extracted.{}.{}", epoch, dynamicOracle);
if (std::filesystem::exists(currentEpochAllExtractedFile))
@@ -51,132 +48,140 @@ void Trainer::extractExamples(SubConfig & config, bool debug, std::filesystem::p
int totalNbExamples = 0;
- while (true)
+ for (auto & config : configs)
{
- if (debug)
- config.printForDebug(stderr);
+ config.addPredicted(machine.getPredicted());
+ config.setStrategy(machine.getStrategyDefinition());
+ config.setState(config.getStrategy().getInitialState());
+ machine.getClassifier(config.getState())->setState(config.getState());
- if (machine.hasSplitWordTransitionSet())
- config.setAppliableSplitTransitions(machine.getSplitWordTransitionSet().getNAppliableTransitions(config, Config::maxNbAppliableSplitTransitions));
+ while (true)
+ {
+ if (debug)
+ config.printForDebug(stderr);
- auto appliableTransitions = machine.getTransitionSet(config.getState()).getAppliableTransitions(config);
- config.setAppliableTransitions(appliableTransitions);
+ if (machine.hasSplitWordTransitionSet())
+ config.setAppliableSplitTransitions(machine.getSplitWordTransitionSet().getNAppliableTransitions(config, Config::maxNbAppliableSplitTransitions));
- std::vector<std::vector<long>> context;
+ auto appliableTransitions = machine.getTransitionSet(config.getState()).getAppliableTransitions(config);
+ config.setAppliableTransitions(appliableTransitions);
- try
- {
- context = machine.getClassifier(config.getState())->getNN()->extractContext(config);
- } catch(std::exception & e)
- {
- util::myThrow(fmt::format("Failed to extract context : {}", e.what()));
- }
+ std::vector<std::vector<long>> context;
- Transition * transition = nullptr;
+ try
+ {
+ context = machine.getClassifier(config.getState())->getNN()->extractContext(config);
+ } catch(std::exception & e)
+ {
+ util::myThrow(fmt::format("Failed to extract context : {}", e.what()));
+ }
- auto goldTransitions = machine.getTransitionSet(config.getState()).getBestAppliableTransitions(config, appliableTransitions, true or dynamicOracle);
+ Transition * transition = nullptr;
- Transition * goldTransition = goldTransitions[0];
- if (config.getState() == "parser")
- goldTransitions[std::rand()%goldTransitions.size()];
+ auto goldTransitions = machine.getTransitionSet(config.getState()).getBestAppliableTransitions(config, appliableTransitions, true or dynamicOracle);
- int nbClasses = machine.getTransitionSet(config.getState()).size();
+ Transition * goldTransition = goldTransitions[0];
+ if (config.getState() == "parser")
+ goldTransitions[std::rand()%goldTransitions.size()];
- float bestScore = -std::numeric_limits<float>::max();
+ int nbClasses = machine.getTransitionSet(config.getState()).size();
- float entropy = 0.0;
-
- if (dynamicOracle and util::choiceWithProbability(1.0) and config.getState() != "tokenizer" and config.getState() != "segmenter")
- {
- auto neuralInput = torch::from_blob(context[0].data(), {(long)context[0].size()}, torch::kLong).clone().to(NeuralNetworkImpl::device);
- auto prediction = torch::softmax(machine.getClassifier(config.getState())->getNN()(neuralInput), -1).squeeze(0);
- entropy = NeuralNetworkImpl::entropy(prediction);
-
- std::vector<int> candidates;
+ float bestScore = -std::numeric_limits<float>::max();
- for (unsigned int i = 0; i < prediction.size(0); i++)
+ float entropy = 0.0;
+
+ if (dynamicOracle and util::choiceWithProbability(1.0) and config.getState() != "tokenizer" and config.getState() != "segmenter")
{
- float score = prediction[i].item<float>();
- if (score > bestScore and appliableTransitions[i])
- bestScore = score;
+ auto neuralInput = torch::from_blob(context[0].data(), {(long)context[0].size()}, torch::kLong).clone().to(NeuralNetworkImpl::device);
+ auto prediction = torch::softmax(machine.getClassifier(config.getState())->getNN()(neuralInput), -1).squeeze(0);
+ entropy = NeuralNetworkImpl::entropy(prediction);
+
+ std::vector<int> candidates;
+
+ for (unsigned int i = 0; i < prediction.size(0); i++)
+ {
+ float score = prediction[i].item<float>();
+ if (score > bestScore and appliableTransitions[i])
+ bestScore = score;
+ }
+
+ for (unsigned int i = 0; i < prediction.size(0); i++)
+ {
+ float score = prediction[i].item<float>();
+ if (appliableTransitions[i] and bestScore - score <= explorationThreshold)
+ candidates.emplace_back(i);
+ }
+
+ transition = machine.getTransitionSet(config.getState()).getTransition(candidates[std::rand()%candidates.size()]);
}
-
- for (unsigned int i = 0; i < prediction.size(0); i++)
+ else
{
- float score = prediction[i].item<float>();
- if (appliableTransitions[i] and bestScore - score <= explorationThreshold)
- candidates.emplace_back(i);
+ transition = goldTransition;
}
- transition = machine.getTransitionSet(config.getState()).getTransition(candidates[std::rand()%candidates.size()]);
- }
- else
- {
- transition = goldTransition;
- }
-
- if (!transition or !goldTransition)
- {
- config.printForDebug(stderr);
- util::myThrow("No transition appliable !");
- }
+ if (!transition or !goldTransition)
+ {
+ config.printForDebug(stderr);
+ util::myThrow("No transition appliable !");
+ }
- std::vector<long> goldIndexes;
- bool exampleIsBanned = machine.getClassifier(config.getState())->exampleIsBanned(config);
+ std::vector<long> goldIndexes;
+ bool exampleIsBanned = machine.getClassifier(config.getState())->exampleIsBanned(config);
- if (machine.getClassifier(config.getState())->isRegression())
- {
- entropy = 0.0;
- auto errMessage = fmt::format("Invalid regression transition '{}'", transition->getName());
- auto splited = util::split(transition->getName(), ' ');
- if (splited.size() != 3 or splited[0] != "WRITESCORE")
- util::myThrow(errMessage);
- auto col = splited[2];
- splited = util::split(splited[1], '.');
- if (splited.size() != 2)
- util::myThrow(errMessage);
- auto object = Config::str2object(splited[0]);
- int index = std::stoi(splited[1]);
-
- float regressionTarget = std::stof(config.getConst(col, config.getRelativeWordIndex(object, index), 0));
- goldIndexes.emplace_back(util::float2long(regressionTarget));
- }
- else
- {
- for (auto & t : goldTransitions)
- goldIndexes.emplace_back(machine.getTransitionSet(config.getState()).getTransitionIndex(t));
+ if (machine.getClassifier(config.getState())->isRegression())
+ {
+ entropy = 0.0;
+ auto errMessage = fmt::format("Invalid regression transition '{}'", transition->getName());
+ auto splited = util::split(transition->getName(), ' ');
+ if (splited.size() != 3 or splited[0] != "WRITESCORE")
+ util::myThrow(errMessage);
+ auto col = splited[2];
+ splited = util::split(splited[1], '.');
+ if (splited.size() != 2)
+ util::myThrow(errMessage);
+ auto object = Config::str2object(splited[0]);
+ int index = std::stoi(splited[1]);
+
+ float regressionTarget = std::stof(config.getConst(col, config.getRelativeWordIndex(object, index), 0));
+ goldIndexes.emplace_back(util::float2long(regressionTarget));
+ }
+ else
+ {
+ for (auto & t : goldTransitions)
+ goldIndexes.emplace_back(machine.getTransitionSet(config.getState()).getTransitionIndex(t));
- }
+ }
- if (!exampleIsBanned)
- {
- totalNbExamples += context.size();
- if (totalNbExamples >= (int)safetyNbExamplesMax)
- util::myThrow(fmt::format("Trying to extract more examples than the limit ({})", util::int2HumanStr(safetyNbExamplesMax)));
+ if (!exampleIsBanned)
+ {
+ totalNbExamples += context.size();
+ if (totalNbExamples >= (int)safetyNbExamplesMax)
+ util::myThrow(fmt::format("Trying to extract more examples than the limit ({})", util::int2HumanStr(safetyNbExamplesMax)));
- examplesPerState[config.getState()].addContext(context);
- examplesPerState[config.getState()].addClass(machine.getClassifier(config.getState())->getLossFunction(), nbClasses, goldIndexes);
- examplesPerState[config.getState()].saveIfNeeded(config.getState(), dir, maxNbExamplesPerFile, epoch, dynamicOracle);
- }
+ examplesPerState[config.getState()].addContext(context);
+ examplesPerState[config.getState()].addClass(machine.getClassifier(config.getState())->getLossFunction(), nbClasses, goldIndexes);
+ examplesPerState[config.getState()].saveIfNeeded(config.getState(), dir, maxNbExamplesPerFile, epoch, dynamicOracle);
+ }
- config.setChosenActionScore(bestScore);
+ config.setChosenActionScore(bestScore);
- transition->apply(config, entropy);
- config.addToHistory(transition->getName());
+ transition->apply(config, entropy);
+ config.addToHistory(transition->getName());
- auto movement = config.getStrategy().getMovement(config, transition->getName());
- if (debug)
- fmt::print(stderr, "(Transition,Newstate,Movement) = ({},{},{})\n", transition->getName(), movement.first, movement.second);
- if (movement == Strategy::endMovement)
- break;
+ auto movement = config.getStrategy().getMovement(config, transition->getName());
+ if (debug)
+ fmt::print(stderr, "(Transition,Newstate,Movement) = ({},{},{})\n", transition->getName(), movement.first, movement.second);
+ if (movement == Strategy::endMovement)
+ break;
- config.setState(movement.first);
- machine.getClassifier(config.getState())->setState(movement.first);
- config.moveWordIndexRelaxed(movement.second);
+ config.setState(movement.first);
+ machine.getClassifier(config.getState())->setState(movement.first);
+ config.moveWordIndexRelaxed(movement.second);
- if (config.needsUpdate())
- config.update();
- }
+ if (config.needsUpdate())
+ config.update();
+ } // End while true
+ } // End for on configs
for (auto & it : examplesPerState)
it.second.saveIfNeeded(it.first, dir, 0, epoch, dynamicOracle);