NarrowLookupTable:[accUpdateGradParameters,type] work (unit tested)

1 files changed, 32 insertions, 2 deletions

diff --git a/test/test-all.lua b/test/test-all.lua
index 28254a7..1d4f617 100644
--- a/test/test-all.lua
+++ b/test/test-all.lua
@@ -515,6 +515,7 @@ function nnxtest.NarrowLookupTable()
    local batchSize = 8
    local embedSize = 32
    local deltaSize = 4
+   local lr = 0.1
    
    -- 1D input
    local input = torch.randperm(dictSize):narrow(1,1,nIndex)
@@ -537,16 +538,29 @@ function nnxtest.NarrowLookupTable()
    local idx = 1
    local narrowSize = embedSize
    for i=1,nIndex do
-      gradWeight2[input[i]]:narrow(1, 1, narrowSize):copy(output:narrow(1,idx,narrowSize))
+      gradWeight2[input[i]]:narrow(1, 1, narrowSize):add(output:narrow(1,idx,narrowSize))
       idx = idx + narrowSize
       narrowSize = narrowSize - deltaSize
    end
    mytester:assertTensorEq(nlt.gradWeight, gradWeight2, 0.000001, "1D backward error")
    
+   nlt:zeroGradParameters()
+   local weight2 = nlt.weight:clone()
+   nlt:backwardUpdate(input, output, lr)
+   local idx = 1
+   local narrowSize = embedSize
+   for i=1,nIndex do
+      weight2[input[i]]:narrow(1, 1, narrowSize):add(-lr, output:narrow(1,idx,narrowSize))
+      idx = idx + narrowSize
+      narrowSize = narrowSize - deltaSize
+   end
+   mytester:assertTensorEq(nlt.weight, weight2, 0.000001, "1D backwardUpdate error")
+   
    -- 2D input
+   nlt:float()
    local input = torch.randperm(dictSize):narrow(1,1,nIndex*batchSize):view(8,-1)
    local output = nlt:forward(input)
-   local output2 = torch.Tensor(batchSize, 120):zero()
+   local output2 = torch.FloatTensor(batchSize, 120):zero()
    for k=1,batchSize do
       local input = input[k]
       local output2 = output2[k]
@@ -575,6 +589,22 @@ function nnxtest.NarrowLookupTable()
       end
    end
    mytester:assertTensorEq(nlt.gradWeight, gradWeight2, 0.000001, "2D backward error")
+   
+   nlt:zeroGradParameters()
+   local weight2 = nlt.weight:clone()
+   nlt:backwardUpdate(input, output, lr)
+   for k=1,batchSize do
+      local input = input[k]
+      local output = output[k]
+      local idx = 1
+      local narrowSize = embedSize
+      for i=1,nIndex do
+         weight2[input[i]]:narrow(1,1,narrowSize):add(-lr, output:narrow(1,idx,narrowSize))
+         idx = idx + narrowSize
+         narrowSize = narrowSize - deltaSize
+      end
+   end
+   mytester:assertTensorEq(nlt.weight, weight2, 0.000001, "2D backwardUpdate error")
 end
 
 function nnx.test(tests)