public static void TrainSpellingModel()
{
string trainfile = AppDomain.CurrentDomain.BaseDirectory + "\\wordvec\\spelling_train_data.txt";
string outputFileName = AppDomain.CurrentDomain.BaseDirectory + "\\wordvec\\spelling_model.bin";
var word2Vec = Word2VecBuilder.Create()
.WithTrainFile(trainfile) // Use text data to train the model;
.WithOutputFile(outputFileName) //Use to save the resulting word vectors / word clusters
.WithSize(200) //Set size of word vectors; default is 100
.WithSaveVocubFile(AppDomain.CurrentDomain.BaseDirectory + "/wordvec/noisy_spelling.txt") //The vocabulary will be saved to <file>
.WithDebug(2) //Set the debug mode (default = 2 = more info during training)
.WithBinary(1) //Save the resulting vectors in binary moded; default is 0 (off)
.WithCBow(1) //Use the continuous bag of words model; default is 1 (use 0 for skip-gram model)
.WithAlpha(0.05f) //Set the starting learning rate; default is 0.025 for skip-gram and 0.05 for CBOW
.WithWindow(10) //Set max skip length between words; default is 5
.WithSample((float)1e-3) //Set threshold for occurrence of words. Those that appear with higher frequency in the training data twill be randomly down-sampled; default is 1e-3, useful range is (0, 1e-5)
.WithHs(0) //Use Hierarchical Softmax; default is 0 (not used)
.WithNegative(5) //Number of negative examples; default is 5, common values are 3 - 10 (0 = not used)
.WithThreads(8) //Use <int> threads (default 12)
.WithIter(5) //Run more training iterations (default 5)
.WithMinCount(1) //This will discard words that appear less than <int> times; default is 5
.WithClasses(0) //Output word classes rather than word vectors; default number of classes is 0 (vectors are written)
.Build();
word2Vec.TrainModel();
}
public void TrainModel()
{
string trainFile = Configuration.TrainFile;
string outputFile = Configuration.OutputFile;
string vocabFile = Configuration.VocabularyFile;
if (!File.Exists(outputFile) && !File.Exists(vocabFile) && File.Exists(trainFile))
{
var word2Vec = Word2VecBuilder.Create()
.WithTrainFile(trainFile)
.WithOutputFile(outputFile)
.WithSize(200)
.WithSaveVocubFile(vocabFile)
.Build();
word2Vec.TrainModel();
}
}
public void TestAppInit()
{
var word2Vec = Word2VecBuilder.Create()
.WithTrainFile("TestData/faust1.txt") // Use text data to train the model;
.WithOutputFile("faust1.bin") //Use to save the resulting word vectors / word clusters
.WithSize(200) //Set size of word vectors; default is 100
.WithSaveVocubFile("faust1.dic") //The vocabulary will be saved to <file>
.WithCBow(1) //Use the continuous bag of words model; default is 1 (use 0 for skip-gram model)
.WithAlpha(0.05f) //Set the starting learning rate; default is 0.025 for skip-gram and 0.05 for CBOW
.WithWindow(7) //Set max skip length between words; default is 5
.WithSample((float)1e-3) //Set threshold for occurrence of words. Those that appear with higher frequency in the training data twill be randomly down-sampled; default is 1e-3, useful range is (0, 1e-5)
.WithHs(0) //Use Hierarchical Softmax; default is 0 (not used)
.WithNegative(5) //Number of negative examples; default is 5, common values are 3 - 10 (0 = not used)
.WithThreads(5) //Use <int> threads (default 12)
.WithIter(5) //Run more training iterations (default 5)
.WithMinCount(5) //This will discard words that appear less than <int> times; default is 5
.WithClasses(0) //Output word classes rather than word vectors; default number of classes is 0 (vectors are written)
.Build();
word2Vec.TrainModel();
}
/// <summary>
/// This part trains model
/// </summary>
/// <author>
/// This method with libraries in Word2Vec.Net-master was programmed by GitHub user Eabdullin
/// link to GitHub: https://github.com/eabdullin/Word2Vec.Net
/// </author>
/// <param name="trainfile">This file is located in SourceTXT folder</param>
public static void TrainModel(string trainfile, int sizeOfVectors = 100, int minCount = 5, int iterations = 5)
{
string outputFileName = DirectoryManager.GetSpecifiedDirectory("Models") + DirectoryManager.sep + Path.GetFileName(trainfile);
var word2Vec = Word2VecBuilder.Create()
.WithTrainFile(trainfile) // Use text data to train the model;
.WithOutputFile(outputFileName) //Use to save the resulting word vectors / word clusters
.WithSize(sizeOfVectors) //Set size of word vectors; default is 100
.WithDebug(2) //Set the debug mode (default = 2 = more info during training)
.WithCBow(1) //Use the continuous bag of words model; default is 1 (use 0 for skip-gram model)
.WithAlpha(0.05f) //Set the starting learning rate; default is 0.025 for skip-gram and 0.05 for CBOW
.WithSample((float)1e-3) //Set threshold for occurrence of words.
.WithHs(0) //Use Hierarchical Softmax; default is 0 (not used)
.WithNegative(5) //Number of negative examples; default is 5, common values are 3 - 10 (0 = not used)
.WithThreads(12) //Use <int> threads (default 12)
.WithIter(iterations) //Run more training iterations (default 5)
.WithMinCount(minCount) //This will discard words that appear less than <int> times; default is 5
.Build();
word2Vec.TrainModel();
}
private void Button_Click(object sender, RoutedEventArgs e)
{
if (t != null && !t.IsCompleted)
{
warnings.Content = "Training in progress, please wait for it to finish";
return;
}
Word2Vec.Net.Word2Vec word2Vec = null;
try
{
word2Vec = Word2VecBuilder.Create()
.WithTrainFile(trainSet.Text) // Use text data to train the model;
.WithOutputFile(@"MODELS\" + outputName.Text + ".bin") //Use to save the resulting word vectors / word clusters
.WithSize(int.Parse(vectorSizeTB.Text)) //Set size of word vectors; default is 100
//.WithSaveVocubFile()//The vocabulary will be saved to <file>
.WithDebug(2) //Set the debug mode (default = 2 = more info during training)
.WithBinary(1) //Save the resulting vectors in binary moded; default is 0 (off)
.WithCBow(modelCB.SelectedIndex) //Use the continuous bag of words model; default is 1 (use 0 for skip-gram model)
.WithAlpha(float.Parse(learningRateTB.Text)) //Set the starting learning rate; default is 0.025 for skip-gram and 0.05 for CBOW
.WithWindow(7) //Set max skip length between words; default is 5
.WithSample((float)1e-3) //Set threshold for occurrence of words. Those that appear with higher frequency in the training data twill be randomly down-sampled; default is 1e-3, useful range is (0, 1e-5)
.WithHs(0) //Use Hierarchical Softmax; default is 0 (not used)
.WithNegative(int.Parse(negtb.Text)) //Number of negative examples; default is 5, common values are 3 - 10 (0 = not used)
.WithThreads(int.Parse(threadsCB.Text)) //Use <int> threads (default 12)
.WithIter(int.Parse(iterTB.Text)) //Run more training iterations (default 5)
.WithMinCount(30) //This will discard words that appear less than <int> times; default is 5
.WithClasses(0) //Output word classes rather than word vectors; default number of classes is 0 (vectors are written)
.Build();
}
catch (Exception e1)
{
warnings.Content = "Wrong input";
return;
}
warnings.Content = "Training started.";
t = new Task(word2Vec.TrainModel);
t.Start();
}
static void Main(string[] args)
{
if (args.Length == 0)
{
Console.WriteLine("WORD VECTOR estimation toolkit v 0.1c\n");
Console.WriteLine("Options:");
Console.WriteLine("Parameters for training:");
Console.WriteLine("\t-train <file>");
Console.WriteLine("\t\tUse text data from <file> to train the model");
Console.WriteLine("\t-output <file>");
Console.WriteLine("\t\tUse <file> to save the resulting word vectors / word clusters");
Console.WriteLine("\t-size <int>");
Console.WriteLine("\t\tSet size of word vectors; default is 100");
Console.WriteLine("\t-window <int>");
Console.WriteLine("\t\tSet max skip length between words; default is 5");
Console.WriteLine("\t-sample <float>");
Console.WriteLine(
"\t\tSet threshold for occurrence of words. Those that appear with higher frequency in the training data");
Console.WriteLine("\t\twill be randomly down-sampled; default is 1e-3, useful range is (0, 1e-5)");
Console.WriteLine("\t-hs <int>");
Console.WriteLine("\t\tUse Hierarchical Softmax; default is 0 (not used)");
Console.WriteLine("\t-negative <int>");
Console.WriteLine(
"\t\tNumber of negative examples; default is 5, common values are 3 - 10 (0 = not used)");
Console.WriteLine("\t-threads <int>");
Console.WriteLine("\t\tUse <int> threads (default 12)");
Console.WriteLine("\t-iter <int>");
Console.WriteLine("\t\tRun more training iterations (default 5)");
Console.WriteLine("\t-min-count <int>");
Console.WriteLine("\t\tThis will discard words that appear less than <int> times; default is 5");
Console.WriteLine("\t-alpha <float>");
Console.WriteLine("\t\tSet the starting learning rate; default is 0.025 for skip-gram and 0.05 for CBOW");
Console.WriteLine("\t-classes <int>");
Console.WriteLine(
"\t\tOutput word classes rather than word vectors; default number of classes is 0 (vectors are written)");
Console.WriteLine("\t-debug <int>");
Console.WriteLine("\t\tSet the debug mode (default = 2 = more info during training)");
Console.WriteLine("\t-binary <int>");
Console.WriteLine("\t\tSave the resulting vectors in binary moded; default is 0 (off)");
Console.WriteLine("\t-save-vocab <file>");
Console.WriteLine("\t\tThe vocabulary will be saved to <file>");
Console.WriteLine("\t-read-vocab <file>");
Console.WriteLine("\t\tThe vocabulary will be read from <file>, not constructed from the training data");
Console.WriteLine("\t-cbow <int>");
Console.WriteLine("\t\tUse the continuous bag of words model; default is 1 (use 0 for skip-gram model)");
Console.WriteLine("Examples:");
Console.WriteLine(
"./word2vec -train data.txt -output vec.txt -size 200 -window 5 -sample 1e-4 -negative 5 -hs 0 -binary 0 -cbow 1 -iter 3");
Console.ReadLine();
return;
}
int i;
var builder = Word2VecBuilder.Create();
string outputFileName = String.Empty;
if ((i = ArgPos("-train", args)) > -1)
{
builder.WithTrainFile(args[i + 1]);
}
if ((i = ArgPos("-output", args)) > -1)
{
outputFileName = args[i + 1];
builder.WithOutputFile(outputFileName);
}
if ((i = ArgPos("-size", args)) > -1)
{
builder.WithSize(int.Parse(args[i + 1]));
}
if ((i = ArgPos("-save-vocab", args)) > -1)
{
builder.WithSaveVocubFile(args[i + 1]);
}
if ((i = ArgPos("-read-vocab", args)) > -1)
{
builder.WithReadVocubFile(args[i + 1]);
}
if ((i = ArgPos("-debug", args)) > -1)
{
builder.WithDebug(int.Parse(args[i + 1]));
}
if ((i = ArgPos("-binary", args)) > -1)
{
builder.WithBinary(int.Parse(args[i + 1]));
}
if ((i = ArgPos("-cbow", args)) > -1)
{
builder.WithCBow(int.Parse(args[i + 1]));
}
if ((i = ArgPos("-alpha", args)) > -1)
{
builder.WithAlpha(float.Parse(args[i + 1]));
}
if ((i = ArgPos("-window", args)) > -1)
{
builder.WithWindow(int.Parse(args[i + 1]));
}
if ((i = ArgPos("-sample", args)) > -1)
{
builder.WithSample(float.Parse(args[i + 1]));
}
if ((i = ArgPos("-hs", args)) > -1)
{
builder.WithHs(int.Parse(args[i + 1]));
}
if ((i = ArgPos("-negative", args)) > -1)
{
builder.WithNegative(int.Parse(args[i + 1]));
}
if ((i = ArgPos("-threads", args)) > -1)
{
builder.WithThreads(int.Parse(args[i + 1]));
}
if ((i = ArgPos("-iter", args)) > -1)
{
builder.WithIter(int.Parse(args[i + 1]));
}
if ((i = ArgPos("-min-count", args)) > -1)
{
builder.WithMinCount(int.Parse(args[i + 1]));
}
if ((i = ArgPos("-classes", args)) > -1)
{
builder.WithClasses(int.Parse(args[i + 1]));
}
Word2Vec.Net.Word2Vec word2Vec = builder.Build();
word2Vec.TrainModel();
var distance = new Distance(outputFileName);
while (true)
{
Console.WriteLine("Distance: Enter word or sentence (EXIT to break): ");
string text = Console.ReadLine();
if (text == null || text.ToLower().Equals("exit"))
{
break;
}
var result = distance.Search(text);
Console.WriteLine("\n Word Cosine distance\n------------------------------------------------------------------------");
foreach (var bestWord in result.Where(x => !string.IsNullOrEmpty(x.Word)))
{
Console.WriteLine("{0}\t\t{1}", bestWord.Word, bestWord.Distance);
}
Console.WriteLine();
}
}
