started working on SoftMaxTree_updateOutput

1 files changed, 161 insertions, 0 deletions

diff --git a/generic/SoftMaxTree.c b/generic/SoftMaxTree.c
new file mode 100644
index 0000000..3967b38
--- /dev/null
+++ b/generic/SoftMaxTree.c
@@ -0,0 +1,161 @@
+#ifndef TH_GENERIC_FILE
+#define TH_GENERIC_FILE "generic/SoftMaxTree.c"
+#else
+
+static int nn_(SoftMaxTree_updateOutput)(lua_State *L)
+{
+  THTensor *input = luaT_checkudata(L, 2, torch_Tensor);  
+  THTensor *target = luaT_checkudata(L, 3, torch_Tensor);  
+  int inputSize = luaT_getfieldcheckint(L, 1, "inputSize");
+  long rootId = (long)(luaT_getfieldcheckint(L, 1, "rootId"));
+  
+  THTensor *weight = luaT_getfieldcheckudata(L, 1, "weight", torch_Tensor);
+  THTensor *bias = luaT_getfieldcheckudata(L, 1, "bias", torch_Tensor);
+  THTensor *output = luaT_getfieldcheckudata(L, 1, "output", torch_Tensor);
+  THTensor *childParent = luaT_getfieldcheckudata(L, 1, "childParent", torch_Tensor);
+  
+  THTensor *node, *nodeWeight, *nodeBias, *nodeOutput;
+  
+
+  long k, i;
+  
+  luaL_argcheck(L, input->nDimension == 2, 2, "2D(batch mode) tensor expected");
+  luaL_argcheck(L, input->size[1] == inputSize, 2, "invalid input size");
+
+  node = THTensor_(new)();
+  nodeWeight = THTensor_(new)();
+  nodeBias = THTensor_(new)();
+  nodeOutput = THTensor_(new)();
+  
+  THTensor_(resize1d)(output, input->size[0]);
+  
+  for(i = 0; i < input->size[0], i++)
+  {
+    long childId = (long)(THTensor_(get1d)(target, i));
+    real activation = 
+
+    while (1)
+    {
+      long parentId = (long)(THTensor_(get1d)(childParent, childId));
+      long parentIdx, nChildren;
+      
+      THTensor_(select)(node, parentChildren, 0, parentId);
+      parentIdx = (long)(THTensor_(get1d)(node, 0));
+      nChildren = (long)(THTensor_(get1d)(node, 1));
+      
+      THTensor_(narrow)(nodeWeight, weight, 0, parentIdx, nChildren);
+      THTensor_(narrow)(nodeBias, bias, 0, parentIdx, nChildren);
+      TH_API void THTensor_(addmv)(THTensor *r_, real beta, THTensor *t, real alpha, THTensor *mat,  THTensor *vec);
+      
+      THTensor_(addmv)(nodeOutput, 1, nodeBias, 1, nodeWeight, input);
+      
+      
+      
+      if parentId == rootId) 
+      {
+        break;
+      }
+      childId = parentId;
+    }
+    end
+    -- sample channel (one channel per sample)
+    local channel = nn.Sequential()
+    channel:add(concat)
+    channel:add(nn.CMulTable())
+    parallel:add(channel)
+  
+  }
+  THTensor_(free)(node);
+  THTensor_(free)(nodeWeight);
+  THTensor_(free)(nodeBias);
+  THTensor_(free)(nodeOutput);
+  return 1;
+}
+
+static int nn_(SoftMaxTree_updateGradInput)(lua_State *L)
+{
+  THTensor *input = luaT_checkudata(L, 2, torch_Tensor);  
+  THTensor *gradOutput = luaT_checkudata(L, 3, torch_Tensor);  
+  int kW = luaT_getfieldcheckint(L, 1, "kW");
+  int dW = luaT_getfieldcheckint(L, 1, "dW");
+  long nInputFrame;
+  long nOutputFrame;
+
+  THTensor *weight = luaT_getfieldcheckudata(L, 1, "weight", torch_Tensor);
+  THTensor *gradInput = luaT_getfieldcheckudata(L, 1, "gradInput", torch_Tensor);
+
+  THTensor *gradOutputWindow;
+  THTensor *gradInputWindow;
+  long k, i;
+  
+  int dimS = 0; // sequence dimension
+  int dimF = 1; // feature dimension
+  
+  if (gradOutput->nDimension == 3) 
+  {
+    dimS = 1;
+    dimF = 2;
+  }
+  
+  nInputFrame = input->size[dimS];
+  nOutputFrame = gradOutput->size[dimS];
+
+  
+
+  return 1;
+}
+
+static int nn_(SoftMaxTree_accGradParameters)(lua_State *L)
+{
+  THTensor *input = luaT_checkudata(L, 2, torch_Tensor);  
+  THTensor *gradOutput = luaT_checkudata(L, 3, torch_Tensor);  
+  real scale = luaL_optnumber(L, 4, 1);
+  int kW = luaT_getfieldcheckint(L, 1, "kW");
+  int dW = luaT_getfieldcheckint(L, 1, "dW");
+  long nInputFrame;
+  long nOutputFrame;
+
+  THTensor *gradWeight = luaT_getfieldcheckudata(L, 1, "gradWeight", torch_Tensor);
+  THTensor *gradBias = luaT_getfieldcheckudata(L, 1, "gradBias", torch_Tensor);
+
+  THTensor *gradOutputWindow;
+  THTensor *inputWindow;
+  long k, i;
+  
+  int dimS = 0; // sequence dimension
+  int dimF = 1; // feature dimension
+  
+  if (gradOutput->nDimension == 3) 
+  {
+    dimS = 1;
+    dimF = 2;
+  }
+  
+  nInputFrame = input->size[dimS];
+  nOutputFrame = gradOutput->size[dimS];
+
+  input = THTensor_(newContiguous)(input);
+  gradOutputWindow = THTensor_(new)();
+  inputWindow = THTensor_(new)();
+  
+  
+  THTensor_(free)(input);
+
+  return 0;
+}
+
+static const struct luaL_Reg nn_(SoftMaxTree__) [] = {
+  {"TemporalConvolution_updateOutput", nn_(SoftMaxTree_updateOutput)},
+  {"TemporalConvolution_updateGradInput", nn_(SoftMaxTree_updateGradInput)},
+  {"TemporalConvolution_accGradParameters", nn_(SoftMaxTree_accGradParameters)},
+  {NULL, NULL}
+};
+
+static void nn_(SoftMaxTree_init)(lua_State *L)
+{
+  luaT_pushmetatable(L, torch_Tensor);
+  luaT_registeratname(L, nn_(SoftMaxTree__), "nn");
+  lua_pop(L,1);
+}
+
+#endif