public static bool Train(string currentDirectory, string searchPattern,
Func <bool> IsTerminated)
{
currentDirectory = @"D:\Mozart\src\";
var lex = new CSharp();
string outputFileName = @"D:\Mozart\Mozart.cbow";
Matrix <Word> Model = null;
if (File.Exists(outputFileName))
{
Model = LoadFromFile(outputFileName,
SIZE, out string fmt, out int dims);
}
else
{
Model = BuildFromPlainText(currentDirectory,
"*.cs", lex, outputFileName);
}
TrainMikolovModel(MakeFileList(new string[] { currentDirectory },
"*.cs", SearchOption.AllDirectories),
lex,
Model,
(loss) => { },
IsTerminated);
CBOW.SaveToFile(
Matrix <Word> .Sort(Model),
"CBOW",
CBOW.DIMS,
outputFileName);
return(false);
}
public static Word[] RunFullCosineSort(IOrthography lex, Matrix <Word> Model, string Q, int max)
{
if (Model == null || string.IsNullOrWhiteSpace(Q))
{
Console.ForegroundColor = ConsoleColor.Yellow;
Console.WriteLine("Model not loaded.\r\n");
Console.ResetColor();
Console.WriteLine("See '--load' command for more info...\r\n");
return(null);
}
float[] Re = new float[CBOW.DIMS];
float norm = 0;
var sign = +1;
foreach (var tok in PlainText.ForEach(Q, 0, Q.Length, 0))
{
string wi = lex.GetKey(tok.TextFragment.Substring(tok.StartIndex, tok.Length));
if (wi == "+")
{
sign = +1;
}
else if (wi == "-")
{
sign = -1;
}
else
{
var vec = Model[wi];
if (vec != null)
{
Debug.Assert(vec.Elements.Length == Re.Length);
for (var j = 0; j < Re.Length; j++)
{
Re[j] += sign * vec.Elements[j].Re;
}
norm++;
}
else
{
Console.ForegroundColor = ConsoleColor.Yellow;
Console.WriteLine($"'{wi}' not found.");
Console.ResetColor();
}
}
}
if (norm > 0)
{
for (var j = 0; j < Re.Length; j++)
{
Re[j] /= (float)norm;
}
}
Word[] output = CBOW.Predict(Model, Re, max);
Array.Sort(output,
(a, b) => Dot.CompareTo(a, b));
Console.WriteLine();
Console.WriteLine(" [" + string.Join(",", Re.Select(re => Math.Round(re, 4)).Take(7)) + "...]");
Console.WriteLine();
int len = 0;
for (int i = output.Length - 1; i >= 0; i--)
{
Word n = output[i];
if (n != null)
{
string str = n.Id;
var it = Model[n.Id];
if (it != null)
{
// if (it.Count > 0) {
// var best = it.ArgMax();
// if (best != null) {
// str = best.Id;
// }
// }
}
if (len + str.Length > 37 /* break like if does not fit */)
{
Console.WriteLine(
output.Length <= 31
? $" {str} : {n.ToString(z: true)}"
: $" {str}");
len = 0;
}
else
{
Console.Write(
output.Length <= 31
? $" {str} : {n.ToString(z: true)}"
: $" {str}");
len += str.Length;
}
}
}
Console.WriteLine();
return(output);
}
