void trainCBOW(){
if(rank == 0){
readVocab();
InitNet();
if (negative > 0){
InitUnigramTable();
}
}
trainModelParallelCBOW();
}
void MonolingualModel::train(const string& training_file) {
cout << "MultiVec-mono" << endl;
config.print();
cout << "Training file: " << training_file << endl;
if (!config.freeze) {
// reads vocab and counts words
readVocab(training_file); // TODO: incremental training
if (config.verbose)
cout << "Total number of words: " << training_words << endl;
}
words_processed = 0;
alpha = config.starting_alpha;
// read file to find out the beginning of each chunk
auto chunks = chunkify(training_file, config.n_threads); // also counts the number of lines
if (!config.freeze)
initNet();
if (config.sent_vector)
initSentWeights();
high_resolution_clock::time_point start = high_resolution_clock::now();
if (config.n_threads == 1) {
trainChunk(training_file, chunks, 0);
} else {
vector<thread> threads;
for (int i = 0; i < config.n_threads; ++i) {
threads.push_back(thread(&MonolingualModel::trainChunk, this,
training_file, chunks, i));
}
for (auto it = threads.begin(); it != threads.end(); ++it) {
it->join();
}
}
high_resolution_clock::time_point end = high_resolution_clock::now();
auto duration = duration_cast<microseconds>(end - start).count();
if (config.verbose)
cout << endl;
cout << "Training time: " << static_cast<float>(duration) / 1000000 << endl;
}
