X-Git-Url: https://www.fleuret.org/cgi-bin/gitweb/gitweb.cgi?p=dagnn.git;a=blobdiff_plain;f=test-dagnn.lua;h=b390a29a5b7b412c6ff019e44f51ef5ef5e2a6de;hp=5b266da3ea9e03a6b837f63022ee65b9eecc4198;hb=HEAD;hpb=e6516772e13dd5424f0a1b7e2063a7417614844c

diff --git a/test-dagnn.lua b/test-dagnn.lua
index 5b266da..b390a29 100755
--- a/test-dagnn.lua
+++ b/test-dagnn.lua
@@ -22,12 +22,17 @@
 require 'torch'
 require 'nn'
 
+-- require 'cunn'
+
 require 'dagnn'
 
-function checkGrad(model, criterion, input, target)
+torch.setdefaulttensortype('torch.DoubleTensor')
+torch.manualSeed(1)
+
+function checkGrad(model, criterion, input, target, epsilon)
    local params, gradParams = model:getParameters()
 
-   local epsilon = 1e-5
+   local epsilon = epsilon or 1e-5
 
    local output = model:forward(input)
    local loss = criterion:forward(output, target)
@@ -36,6 +41,8 @@ function checkGrad(model, criterion, input, target)
    model:backward(input, gradOutput)
    local analyticalGradParam = gradParams:clone()
 
+   local err = 0
+
    for i = 1, params:size(1) do
       local x = params[i]
 
@@ -51,23 +58,13 @@ function checkGrad(model, criterion, input, target)
 
       local ana = analyticalGradParam[i]
       local num = (loss1 - loss0) / (2 * epsilon)
-      local err
 
-      if num == ana then
-         err = 0
-      else
-         err = torch.abs(num - ana) / torch.abs(num)
+      if num ~= ana then
+         err = math.max(err, math.abs(num - ana) / math.max(epsilon, math.abs(num)))
       end
-
-      print(
-         'CHECK '
-            .. err
-            .. ' checkGrad ' .. i
-            .. ' analytical ' .. ana
-            .. ' numerical ' .. num
-      )
    end
 
+   return err
 end
 
 function printTensorTable(t)
@@ -81,62 +78,66 @@ function printTensorTable(t)
    end
 end
 
--- torch.setnumthreads(params.nbThreads)
-torch.setdefaulttensortype('torch.DoubleTensor')
-torch.manualSeed(2)
+--               +-- Linear(10, 10) --> ReLU --> d -->
+--              /                               /
+--             /                               /
+--  --> a --> b -----------> c ---------------+
+--                            \
+--                             \
+--                              +--------------- e -->
 
---                     +--> c ----> e --+
---                    /            /     \
---                   /            /       \
---  input --> a --> b ---> d ----+         g --> output
---                          \             /
---                           \           /
---                            +--> f ---+
+dag = nn.DAG()
 
-a = nn.Linear(10, 10)
+a = nn.Linear(50, 10)
 b = nn.ReLU()
-c = nn.Linear(10, 3)
-d = nn.Linear(10, 3)
-e = nn.CMulTable()
-f = nn.Linear(3, 3)
-g = nn.CAddTable()
-
-model = nn.DAG()
-
-model:addEdge(a, b)
-model:addEdge(b, c)
-model:addEdge(b, d)
-model:addEdge(c, e)
-model:addEdge(d, e)
-model:addEdge(d, f)
-model:addEdge(e, g)
-model:addEdge(f, g)
-
-model:setInput(a)
-model:setOutput(g)
-
-input = torch.Tensor(3, 10):uniform()
-
-print('******************************************************************')
-print('** updateOutput **************************************************')
-print('******************************************************************')
-
-output = model:updateOutput(input):clone()
-
-printTensorTable(output)
-
-print('******************************************************************')
-print('** updateGradInput ***********************************************')
-print('******************************************************************')
-
-gradInput = model:updateGradInput(input, output)
+c = nn.Linear(10, 15)
+d = nn.CMulTable()
+e = nn.Mul(-1)
+
+dag:connect(a, b, c)
+dag:connect(b, nn.Linear(10, 15), nn.ReLU(), d)
+dag:connect(c, d)
+dag:connect(c, e)
+
+dag:setLabel(a, 'first module')
+
+dag:setInput(a)
+dag:setOutput({ d, e })
+
+-- Check the output of the dot file. Generate a pdf with:
+--
+-- dot ./graph.dot -Lg -T pdf -o ./graph.pdf
+--
+print('Writing ./graph.dot')
+dag:saveDot('./graph.dot')
+
+-- Let's make a model where the dag is inside another nn.Container.
+model = nn.Sequential()
+   :add(nn.Linear(50, 50))
+   :add(dag)
+   :add(nn.CAddTable())
+
+criterion = nn.MSECriterion()
+
+if cunn then
+   print("Using CUDA")
+   model:cuda()
+   criterion:cuda()
+   torch.setdefaulttensortype('torch.CudaTensor')
+   epsilon = 1e-3
+end
 
-printTensorTable(gradInput)
+local input = torch.Tensor(30, 50):uniform()
+local output = model:updateOutput(input):clone()
+output:uniform()
 
-print('******************************************************************')
-print('** checkGrad *****************************************************')
-print('******************************************************************')
+-- Check that DAG:accGradParameters and friends work okay
+print('Gradient estimate error ' .. checkGrad(model, criterion, input, output, epsilon))
 
-output:uniform()
+-- Check that we can save and reload the model
+model:clearState()
+torch.save('./test.t7', model)
+local otherModel = torch.load('./test.t7')
+print('Gradient estimate error ' .. checkGrad(otherModel, criterion, input, output, epsilon))
 
-checkGrad(model, nn.MSECriterion(), input, output)
+dag:print()