public boolean train(DataIter trainData, boolean cachedLabels, boolean collectIds)
throws Exception {
// map the y-values in the training set.
boolean labelsMapped = false;
if (cachedLabels) {
labelsMapped = stateMappings(trainData);
}
if (dict != null) dict.train(trainData, model.numStates());
for (WordsInTrain d : otherDicts) {
d.train(trainData, model.numStates());
}
boolean requiresTraining = false;
for (int f = 0; f < features.size(); f++) {
if (getFeature(f).requiresTraining()) {
requiresTraining = true;
break;
}
}
if (requiresTraining) {
for (trainData.startScan(); trainData.hasNext(); ) {
DataSequence seq = trainData.next();
for (int f = 0; f < features.size(); f++) {
if (getFeature(f).requiresTraining()) {
trainFeatureType(getFeature(f), seq);
}
}
}
}
if (collectIds) totalFeatures = featureMap.collectFeatureIdentifiers(trainData, maxMemory());
return labelsMapped;
};
public void collectAndFreezeFeatures(DataIter trainData) throws Exception {
featureMap.collectFeatureIdentifiers(trainData, 1);
}
