mirror of
https://github.com/moses-smt/mosesdecoder.git
synced 2025-01-08 12:36:35 +03:00
f658840f1a
git-svn-id: http://svn.statmt.org/repository/mira@3922 cc96ff50-19ce-11e0-b349-13d7f0bd23df
314 lines
13 KiB
C++
314 lines
13 KiB
C++
#include "Optimiser.h"
|
|
#include "Hildreth.h"
|
|
#include "StaticData.h"
|
|
|
|
using namespace Moses;
|
|
using namespace std;
|
|
|
|
namespace Mira {
|
|
|
|
size_t MiraOptimiser::updateWeights(ScoreComponentCollection& currWeights,
|
|
const vector<vector<ScoreComponentCollection> >& featureValues,
|
|
const vector<vector<float> >& losses,
|
|
const vector<vector<float> >& bleuScores,
|
|
const vector<ScoreComponentCollection>& oracleFeatureValues,
|
|
const vector<float> oracleBleuScores,
|
|
const vector<size_t> sentenceIds,
|
|
float learning_rate,
|
|
size_t rank,
|
|
size_t epoch) {
|
|
|
|
// vector of feature values differences for all created constraints
|
|
vector<ScoreComponentCollection> featureValueDiffs;
|
|
vector<float> lossMinusModelScoreDiffs;
|
|
vector<float> all_losses;
|
|
|
|
// most violated constraint in batch
|
|
ScoreComponentCollection max_batch_featureValueDiff;
|
|
float max_batch_lossMinusModelScoreDiff = -1;
|
|
|
|
// Make constraints for new hypothesis translations
|
|
float epsilon = 0.0001;
|
|
int violatedConstraintsBefore = 0;
|
|
float oldDistanceFromOptimum = 0;
|
|
// iterate over input sentences (1 (online) or more (batch))
|
|
for (size_t i = 0; i < featureValues.size(); ++i) {
|
|
//size_t sentenceId = sentenceIds[i];
|
|
// iterate over hypothesis translations for one input sentence
|
|
for (size_t j = 0; j < featureValues[i].size(); ++j) {
|
|
ScoreComponentCollection featureValueDiff = oracleFeatureValues[i];
|
|
featureValueDiff.MinusEquals(featureValues[i][j]);
|
|
|
|
cerr << "Rank " << rank << ", epoch " << epoch << ", feature value diff: " << featureValueDiff << endl;
|
|
if (featureValueDiff.GetL1Norm() == 0) {
|
|
// skip constraint
|
|
continue;
|
|
}
|
|
|
|
float loss = losses[i][j];
|
|
if (m_scale_margin == 1) {
|
|
loss *= oracleBleuScores[i];
|
|
VERBOSE(1, "Rank " << rank << ", epoch " << epoch << ", scaling margin with oracle bleu score " << oracleBleuScores[i] << endl);
|
|
}
|
|
else if (m_scale_margin == 2) {
|
|
loss *= log2(oracleBleuScores[i]);
|
|
VERBOSE(1, "Rank " << rank << ", epoch " << epoch << ", scaling margin with log2 oracle bleu score " << log2(oracleBleuScores[i]) << endl);
|
|
}
|
|
else if (m_scale_margin == 10) {
|
|
loss *= log10(oracleBleuScores[i]);
|
|
VERBOSE(1, "Rank " << rank << ", epoch " << epoch << ", scaling margin with log10 oracle bleu score " << log10(oracleBleuScores[i]) << endl)
|
|
}
|
|
|
|
// check if constraint is violated
|
|
bool violated = false;
|
|
bool addConstraint = true;
|
|
float modelScoreDiff = featureValueDiff.InnerProduct(currWeights);
|
|
float diff = 0;
|
|
if (loss > (modelScoreDiff + m_margin_slack)) {
|
|
diff = loss - (modelScoreDiff + m_margin_slack);
|
|
}
|
|
cerr << "Rank " << rank << ", epoch " << epoch << ", constraint: " << modelScoreDiff << " + " << m_margin_slack << " >= " << loss << " (current violation: " << diff << ")" << endl;
|
|
|
|
if (diff > epsilon) {
|
|
violated = true;
|
|
}
|
|
else if (m_onlyViolatedConstraints) {
|
|
addConstraint = false;
|
|
}
|
|
|
|
float lossMinusModelScoreDiff = loss - modelScoreDiff;
|
|
if (addConstraint) {
|
|
featureValueDiffs.push_back(featureValueDiff);
|
|
lossMinusModelScoreDiffs.push_back(lossMinusModelScoreDiff);
|
|
all_losses.push_back(loss);
|
|
|
|
if (violated) {
|
|
++violatedConstraintsBefore;
|
|
oldDistanceFromOptimum += diff;
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
// run optimisation: compute alphas for all given constraints
|
|
vector<float> alphas;
|
|
ScoreComponentCollection summedUpdate;
|
|
if (violatedConstraintsBefore > 0) {
|
|
cerr << "Rank " << rank << ", epoch " << epoch << ", number of constraints passed to optimizer: " <<
|
|
featureValueDiffs.size() << " (of which violated: " << violatedConstraintsBefore << ")" << endl;
|
|
if (m_slack != 0) {
|
|
alphas = Hildreth::optimise(featureValueDiffs, lossMinusModelScoreDiffs, m_slack);
|
|
} else {
|
|
alphas = Hildreth::optimise(featureValueDiffs, lossMinusModelScoreDiffs);
|
|
}
|
|
|
|
// Update the weight vector according to the alphas and the feature value differences
|
|
// * w' = w' + SUM alpha_i * (h_i(oracle) - h_i(hypothesis))
|
|
for (size_t k = 0; k < featureValueDiffs.size(); ++k) {
|
|
float alpha = alphas[k];
|
|
VERBOSE(1, "Rank " << rank << ", epoch " << epoch << ", alpha: " << alpha << endl);
|
|
ScoreComponentCollection update(featureValueDiffs[k]);
|
|
update.MultiplyEquals(alpha);
|
|
|
|
// sum updates
|
|
summedUpdate.PlusEquals(update);
|
|
}
|
|
}
|
|
else {
|
|
cerr << "Rank " << rank << ", epoch " << epoch << ", no constraint violated for this batch" << endl;
|
|
return 0;
|
|
}
|
|
|
|
// apply learning rate
|
|
if (learning_rate != 1) {
|
|
VERBOSE(1, "Rank " << rank << ", epoch " << epoch << ", update before applying learning rate: " << summedUpdate << endl);
|
|
summedUpdate.MultiplyEquals(learning_rate);
|
|
}
|
|
|
|
// scale update by BLEU of oracle
|
|
if (oracleBleuScores.size() == 1 && m_scale_update) {
|
|
VERBOSE(1, "Rank " << rank << ", epoch " << epoch << ", scaling summed update with log10 oracle bleu score " << log10(oracleBleuScores[0]) << endl);
|
|
summedUpdate.MultiplyEquals(log10(oracleBleuScores[0]));
|
|
}
|
|
|
|
cerr << "Rank " << rank << ", epoch " << epoch << ", update: " << summedUpdate << endl;
|
|
|
|
// apply update to weight vector
|
|
VERBOSE(1, "Rank " << rank << ", epoch " << epoch << ", weights before update: " << currWeights << endl);
|
|
currWeights.PlusEquals(summedUpdate);
|
|
VERBOSE(1, "Rank " << rank << ", epoch " << epoch << ", weights after update: " << currWeights << endl);
|
|
|
|
// Sanity check: are there still violated constraints after optimisation?
|
|
int violatedConstraintsAfter = 0;
|
|
float newDistanceFromOptimum = 0;
|
|
for (size_t i = 0; i < featureValueDiffs.size(); ++i) {
|
|
float modelScoreDiff = featureValueDiffs[i].InnerProduct(currWeights);
|
|
float loss = all_losses[i];
|
|
float diff = loss - (modelScoreDiff + m_margin_slack);
|
|
if (diff > epsilon) {
|
|
++violatedConstraintsAfter;
|
|
newDistanceFromOptimum += diff;
|
|
}
|
|
}
|
|
VERBOSE(1, "Rank " << rank << ", epoch " << epoch << ", violated constraint before: " << violatedConstraintsBefore << ", after: " << violatedConstraintsAfter << ", change: " << violatedConstraintsBefore - violatedConstraintsAfter << endl);
|
|
VERBOSE(1, "Rank " << rank << ", epoch " << epoch << ", error before: " << oldDistanceFromOptimum << ", after: " << newDistanceFromOptimum << ", change: " << oldDistanceFromOptimum - newDistanceFromOptimum << endl);
|
|
return violatedConstraintsAfter;
|
|
}
|
|
|
|
size_t MiraOptimiser::updateWeightsHopeFear(Moses::ScoreComponentCollection& currWeights,
|
|
const std::vector< std::vector<Moses::ScoreComponentCollection> >& featureValuesHope,
|
|
const std::vector< std::vector<Moses::ScoreComponentCollection> >& featureValuesFear,
|
|
const std::vector<std::vector<float> >& bleuScoresHope,
|
|
const std::vector<std::vector<float> >& bleuScoresFear,
|
|
const std::vector< size_t> sentenceIds,
|
|
float learning_rate,
|
|
size_t rank,
|
|
size_t epoch) {
|
|
|
|
// vector of feature values differences for all created constraints
|
|
vector<ScoreComponentCollection> featureValueDiffs;
|
|
vector<float> lossMinusModelScoreDiffs;
|
|
vector<float> all_losses;
|
|
|
|
// most violated constraint in batch
|
|
ScoreComponentCollection max_batch_featureValueDiff;
|
|
float max_batch_lossMinusModelScoreDiff = -1;
|
|
|
|
// Make constraints for new hypothesis translations
|
|
float epsilon = 0.0001;
|
|
int violatedConstraintsBefore = 0;
|
|
float oldDistanceFromOptimum = 0;
|
|
|
|
// iterate over input sentences (1 (online) or more (batch))
|
|
for (size_t i = 0; i < featureValuesHope.size(); ++i) {
|
|
size_t sentenceId = sentenceIds[i]; // keep sentenceId for storing more than 1 oracle..
|
|
// Pair all hope translations with all fear translations for one input sentence
|
|
for (size_t j = 0; j < featureValuesHope[i].size(); ++j) {
|
|
for (size_t k = 0; k < featureValuesFear[i].size(); ++k) {
|
|
ScoreComponentCollection featureValueDiff = featureValuesHope[i][j];
|
|
featureValueDiff.MinusEquals(featureValuesFear[i][k]);
|
|
cerr << "Rank " << rank << ", epoch " << epoch << ", feature value diff: " << featureValueDiff << endl;
|
|
if (featureValueDiff.GetL1Norm() == 0) {
|
|
// skip constraint
|
|
continue;
|
|
}
|
|
|
|
float loss = bleuScoresHope[i][j] - bleuScoresFear[i][k];
|
|
if (m_scale_margin == 1) {
|
|
loss *= bleuScoresHope[i][j];
|
|
VERBOSE(1, "Rank " << rank << ", epoch " << epoch << ", scaling margin with oracle bleu score " << bleuScoresHope[i][j] << endl);
|
|
}
|
|
else if (m_scale_margin == 2) {
|
|
loss *= log2(bleuScoresHope[i][j]);
|
|
VERBOSE(1, "Rank " << rank << ", epoch " << epoch << ", scaling margin with log2 oracle bleu score " << log2(bleuScoresHope[i][j]) << endl);
|
|
}
|
|
else if (m_scale_margin == 10) {
|
|
loss *= log10(bleuScoresHope[i][j]);
|
|
VERBOSE(1, "Rank " << rank << ", epoch " << epoch << ", scaling margin with log10 oracle bleu score " << log10(bleuScoresHope[i][j]) << endl);
|
|
}
|
|
|
|
// check if constraint is violated
|
|
bool violated = false;
|
|
bool addConstraint = true;
|
|
float modelScoreDiff = featureValueDiff.InnerProduct(currWeights);
|
|
float diff = 0;
|
|
if (loss > (modelScoreDiff + m_margin_slack)) {
|
|
diff = loss - (modelScoreDiff + m_margin_slack);
|
|
}
|
|
cerr << "Rank " << rank << ", epoch " << epoch << ", constraint: " << modelScoreDiff << " + " << m_margin_slack << " >= " << loss << " (current violation: " << diff << ")" << endl;
|
|
|
|
if (diff > epsilon) {
|
|
violated = true;
|
|
}
|
|
else if (m_onlyViolatedConstraints) {
|
|
addConstraint = false;
|
|
}
|
|
|
|
float lossMinusModelScoreDiff = loss - modelScoreDiff;
|
|
if (addConstraint) {
|
|
featureValueDiffs.push_back(featureValueDiff);
|
|
lossMinusModelScoreDiffs.push_back(lossMinusModelScoreDiff);
|
|
all_losses.push_back(loss);
|
|
|
|
if (violated) {
|
|
++violatedConstraintsBefore;
|
|
oldDistanceFromOptimum += diff;
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
// run optimisation: compute alphas for all given constraints
|
|
vector<float> alphas;
|
|
ScoreComponentCollection summedUpdate;
|
|
if (violatedConstraintsBefore > 0) {
|
|
cerr << "Rank " << rank << ", epoch " << epoch << ", number of constraints passed to optimizer: " <<
|
|
featureValueDiffs.size() << " (of which violated: " << violatedConstraintsBefore << ")" << endl;
|
|
if (m_slack != 0) {
|
|
alphas = Hildreth::optimise(featureValueDiffs, lossMinusModelScoreDiffs, m_slack);
|
|
} else {
|
|
alphas = Hildreth::optimise(featureValueDiffs, lossMinusModelScoreDiffs);
|
|
}
|
|
|
|
// Update the weight vector according to the alphas and the feature value differences
|
|
// * w' = w' + SUM alpha_i * (h_i(oracle) - h_i(hypothesis))
|
|
for (size_t k = 0; k < featureValueDiffs.size(); ++k) {
|
|
float alpha = alphas[k];
|
|
VERBOSE(1, "Rank " << rank << ", epoch " << epoch << ", alpha: " << alpha << endl);
|
|
ScoreComponentCollection update(featureValueDiffs[k]);
|
|
update.MultiplyEquals(alpha);
|
|
|
|
// scale update by BLEU of hope translation (only two cases defined at the moment)
|
|
if (featureValuesHope.size() == 1 && m_scale_update) { // only defined for batch size 1)
|
|
if (featureValuesHope[0].size() == 1) {
|
|
VERBOSE(1, "Rank " << rank << ", epoch " << epoch << ", scaling update with log10 oracle bleu score " << log10(bleuScoresHope[0][0]) << endl); // only 1 oracle
|
|
update.MultiplyEquals(log10(bleuScoresHope[0][0]));
|
|
} else if (featureValuesFear[0].size() == 1) {
|
|
VERBOSE(1, "Rank " << rank << ", epoch " << epoch << ", scaling update with log10 oracle bleu score " << log10(bleuScoresHope[0][k]) << endl); // k oracles
|
|
update.MultiplyEquals(log10(bleuScoresHope[0][k]));
|
|
}
|
|
}
|
|
|
|
// sum up update
|
|
summedUpdate.PlusEquals(update);
|
|
}
|
|
}
|
|
else {
|
|
cerr << "Rank " << rank << ", epoch " << epoch << ", check, no constraint violated for this batch" << endl;
|
|
return 0;
|
|
}
|
|
|
|
// apply learning rate
|
|
if (learning_rate != 1) {
|
|
VERBOSE(1, "Rank " << rank << ", epoch " << epoch << ", update before applying learning rate: " << summedUpdate << endl);
|
|
summedUpdate.MultiplyEquals(learning_rate);
|
|
}
|
|
|
|
cerr << "Rank " << rank << ", epoch " << epoch << ", update: " << summedUpdate << endl;
|
|
|
|
// apply update to weight vector
|
|
VERBOSE(1, "Rank " << rank << ", epoch " << epoch << ", weights before update: " << currWeights << endl);
|
|
currWeights.PlusEquals(summedUpdate);
|
|
VERBOSE(1, "Rank " << rank << ", epoch " << epoch << ", weights after update: " << currWeights << endl);
|
|
|
|
// Sanity check: are there still violated constraints after optimisation?
|
|
int violatedConstraintsAfter = 0;
|
|
float newDistanceFromOptimum = 0;
|
|
for (size_t i = 0; i < featureValueDiffs.size(); ++i) {
|
|
float modelScoreDiff = featureValueDiffs[i].InnerProduct(currWeights);
|
|
float loss = all_losses[i];
|
|
float diff = loss - (modelScoreDiff + m_margin_slack);
|
|
if (diff > epsilon) {
|
|
++violatedConstraintsAfter;
|
|
newDistanceFromOptimum += diff;
|
|
}
|
|
}
|
|
VERBOSE(1, "Rank " << rank << ", epoch " << epoch << ", check, violated constraint before: " << violatedConstraintsBefore << ", after: " << violatedConstraintsAfter << ", change: " << violatedConstraintsBefore - violatedConstraintsAfter << endl);
|
|
VERBOSE(1, "Rank " << rank << ", epoch " << epoch << ", check, error before: " << oldDistanceFromOptimum << ", after: " << newDistanceFromOptimum << ", change: " << oldDistanceFromOptimum - newDistanceFromOptimum << endl);
|
|
return violatedConstraintsAfter;
|
|
}
|
|
|
|
}
|
|
|