new code of inputDec tensor

source/sample/transformer/T2TModel.cpp

@@ -362,8 +362,8 @@ void T2TModel::MakeMTMaskDec(XTensor &inputEnc, XTensor &inputDec,
     dims[inputDec.order + 1] = len;
     InitTensor(&maskDec, inputDec.order + 2, dims, X_FLOAT, 1.0F, paddingDec.devID, paddingDec.mem);
     
-    /* an upper triangular matrix where the cells of the upper triangular are set to -1e-9.
-     this matrix can be used to prevent the attention to current or following words in
+    /* An upper triangular matrix where the cells of the upper triangular are set to -1e-9.
+       This matrix can be used to block the attention to current or following words in
        a given sequence. */
     _SetDataLowTri(&maskDec, 1e9F, 0);
     _ScaleAndShiftMe(&maskDec, 1.0F, -1e9F);

source/sample/transformer/T2TPredictor.cpp

@@ -118,6 +118,8 @@ predict the next state
 */
 void T2TPredictor::Predict(T2TStateBundle * next, XTensor * encoding, XTensor * inputEnc, XTensor * paddingEnc)
 {
+    int dims[MAX_TENSOR_DIM_NUM];
+
     next->layersEnc.Clear();
     next->layersDec.Clear();
     
@@ -128,10 +130,25 @@ void T2TPredictor::Predict(T2TStateBundle * next, XTensor * encoding, XTensor *
 
     /* word indices of positions up to next state */
     XTensor &inputDec = *NewTensor();
+
+    /* a dummy word that used to as a placeholder when we process the next work */
+    XTensor dummy;
+        
+    for(int i = 0; i < inputEnc->order - 1; i++)
+        dims[i] = inputEnc->GetDim(i);
+    dims[inputEnc->order - 1] = 1;
+
+    InitTensor(&dummy, inputEnc->order, dims, X_INT, 1.0F, inputEnc->devID, inputEnc->mem);
+    dummy.SetZeroAll();
+
+    
+    /* add a new word into the input sequence of the decoder side */
     if(inputLast == NULL)
-        inputDec = s->prediction;
-    else
-        inputDec = Concatenate(*inputLast, s->prediction, inputLast->GetDim(-1));
+        inputDec = Identity(dummy);
+    else{
+        XTensor inputDecSlide = SelectRange(*inputLast, inputLast->order - 1, 0, inputLast->GetDim(-1) - 2);
+        inputDec = Concatenate(inputDecSlide, dummy, inputDecSlide.order - 1);
+    }
 
     /* prediction probabilities */
     XTensor &output = next->prob;
@@ -156,7 +173,6 @@ void T2TPredictor::Predict(T2TStateBundle * next, XTensor * encoding, XTensor *
 
     CheckNTErrors(decoding.order >= 2, "The tensor must be of order 2 or larger!");
 
-    int dims[MAX_TENSOR_DIM_NUM];
     for(int i = 0; i < decoding.order - 1; i++)
         dims[i] = decoding.GetDim(i);
     dims[decoding.order - 2] = 1;