public Layer_Output(DNN dnn_model)
{
foreach (NeuralLink neurallink in dnn_model.neurallinks)
{
if (neurallink.Nt == N_Type.Fully_Connected)
{
layerPooling.Add(new float[1]);
layerMaxPooling_Index.Add(new int[1]);
}
else
{
layerPooling.Add(new float[dnn_model.LFE.MAX_TOKEN_NUM * neurallink.Neural_Out.Number]);
layerMaxPooling_Index.Add(new int[neurallink.Neural_Out.Number]);
}
layerPoolingSecondary.Add(new float[dnn_model.MaxPoolSentenceNumber * neurallink.Neural_Out.Number]);
layerMaxPooling_IndexSecondary.Add(new int[neurallink.Neural_Out.Number]);
layerOutputs.Add(new float[neurallink.Neural_Out.Number]);
LayerDim.Add(neurallink.Neural_Out.Number);
}
Layer_TOP = layerOutputs.Count - 1;
}
public static void Embedding(string inTgtModel, string inTgtVocab, ModelType inTgtModelType, int inTgtMaxRetainedSeqLength,
string inFilename, string outFilenamePrefix, int dim, FeatureList featureList)
{
Dictionary <string, Dictionary <int, float> > tgtDicWord2Vec = new Dictionary <string, Dictionary <int, float> >();
int numDic = 0;
string line = "";
StreamWriter tfeafp = new StreamWriter(outFilenamePrefix + ".words");
DNN tgt_dssm = new DNN(inTgtModel, inTgtModelType, inTgtVocab, inTgtMaxRetainedSeqLength);
if (inTgtModelType == ModelType.DSSM)
{
inTgtMaxRetainedSeqLength = 1;
}
using (StreamReader sr = new StreamReader(inFilename))
{
while ((line = sr.ReadLine()) != null)
{
if (line != "")
{
numDic++;
}
}
}
StreamReader inFile = new StreamReader(inFilename);
//string line = "";
int linecnt = 0;
tfeafp.WriteLine(numDic + " " + dim);
while ((line = inFile.ReadLine()) != null)
{
linecnt++;
if (0 == linecnt % 1000000)
{
Console.Write("|");
}
else if (0 == linecnt % 100000)
{
Console.Write(".");
}
string tgtstr = line;
List <float> tgtvec = null;
tgtvec = tgt_dssm.Forward(tgtstr, tgtDicWord2Vec, featureList);
tfeafp.Write(tgtstr + " ");
foreach (float x in tgtvec)
{
tfeafp.Write(string.Format("{0:0.######} ", x));
}
tfeafp.WriteLine();
}
if (tfeafp != null)
{
tfeafp.Close();
}
Console.WriteLine("total {0} lines processed!", linecnt);
}