X-Git-Url: https://www.fleuret.org/cgi-bin/gitweb/gitweb.cgi?p=dagnn.git;a=blobdiff_plain;f=test-dagnn.lua;h=5d8a309ce9769b547de0fa104b4b5f0b99157fe1;hp=32eed5724f5ea84c8e55fe966b54e53021013902;hb=59490d9da93f28283c79d44a5e7c791cbd623b24;hpb=34ed0d49d9b6b03811cd92c9513edf4ec5d4d2d2

diff --git a/test-dagnn.lua b/test-dagnn.lua
index 32eed57..5d8a309 100755
--- a/test-dagnn.lua
+++ b/test-dagnn.lua
@@ -1,63 +1,123 @@
 #!/usr/bin/env luajit
 
+--[[
+
+   Copyright (c) 2016 Idiap Research Institute, http://www.idiap.ch/
+   Written by Francois Fleuret <francois.fleuret@idiap.ch>
+
+   This file is free software: you can redistribute it and/or modify
+   it under the terms of the GNU General Public License version 3 as
+   published by the Free Software Foundation.
+
+   It is distributed in the hope that it will be useful, but WITHOUT
+   ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
+   or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public
+   License for more details.
+
+   You should have received a copy of the GNU General Public License
+   along with this file.  If not, see <http://www.gnu.org/licenses/>.
+
+]]--
+
 require 'torch'
 require 'nn'
-
 require 'dagnn'
 
-function printTensorTable(t)
-   if torch.type(t) == 'table' then
-      for i, t in pairs(t) do
-         print('-- ELEMENT [' .. i .. '] --')
-         printTensorTable(t)
-      end
-   else
-      print(tostring(t))
-   end
-end
-
--- torch.setnumthreads(params.nbThreads)
 torch.setdefaulttensortype('torch.DoubleTensor')
-torch.manualSeed(2)
+torch.manualSeed(1)
 
-a = nn.Linear(10, 10)
-b = nn.ReLU()
-c = nn.Linear(10, 3)
-d = nn.Linear(10, 3)
-e = nn.CMulTable()
-f = nn.Linear(3, 2)
+function checkGrad(model, criterion, input, target)
+   local params, gradParams = model:getParameters()
+
+   local epsilon = 1e-5
+
+   local output = model:forward(input)
+   local loss = criterion:forward(output, target)
+   local gradOutput = criterion:backward(output, target)
+   gradParams:zero()
+   model:backward(input, gradOutput)
+   local analyticalGradParam = gradParams:clone()
 
---   a -----> b ---> c ----> e ---
---             \           /
---              \--> d ---/
---                    \
---                     \---> f ---
+   local err = 0
 
-g = nn.DAG()
+   for i = 1, params:size(1) do
+      local x = params[i]
 
-g:addEdge(c, e)
-g:addEdge(a, b)
-g:addEdge(d, e)
-g:addEdge(b, c)
-g:addEdge(b, d)
-g:addEdge(d, f)
+      params[i] = x - epsilon
+      local output0 = model:forward(input)
+      local loss0 = criterion:forward(output0, target)
 
-g:setInput({{a}})
-g:setOutput({ e, f })
+      params[i] = x + epsilon
+      local output1 = model:forward(input)
+      local loss1 = criterion:forward(output1, target)
 
-g:print()
+      params[i] = x
 
-input = torch.Tensor(3, 10):uniform()
+      local ana = analyticalGradParam[i]
+      local num = (loss1 - loss0) / (2 * epsilon)
 
-output = g:updateOutput({{ input }})
+      if num ~= ana then
+         err = math.max(err, math.abs(num - ana) / math.abs(num))
+      end
+   end
 
-printTensorTable(output)
+   return err
+end
+
+function printTensorTable(t)
+   if torch.type(t) == 'table' then
+      for i, t in pairs(t) do
+         print('-- ELEMENT [' .. i .. '] --')
+         printTensorTable(t)
+      end
+   else
+      print(tostring(t))
+   end
+end
 
-----------------------------------------------------------------------
+--               +-- Linear(10, 10) --> ReLU --> d -->
+--              /                               /
+--             /                               /
+--  --> a --> b -----------> c ---------------+
+--                            \
+--                             \
+--                              +--------------- e -->
 
-print('******************************************************************')
-print('** updateGradInput ***********************************************')
-print('******************************************************************')
-gradInput = g:updateGradInput({ input }, output)
+dag = nn.DAG()
 
-printTensorTable(gradInput)
+a = nn.Linear(50, 10)
+b = nn.ReLU()
+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:setInput(a)
+dag:setOutput({ d, e })
+
+-- Check the output of the dot file
+print('Writing /tmp/graph.dot')
+dag:saveDot('/tmp/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())
+
+local input = torch.Tensor(30, 50):uniform()
+local output = model:updateOutput(input):clone()
+output:uniform()
+
+-- Check that DAG:accGradParameters and friends work okay
+print('Gradient estimate error ' .. checkGrad(model, nn.MSECriterion(), input, output))
+
+-- Check that we can save and reload the model
+model:clearState()
+torch.save('/tmp/test.t7', model)
+local otherModel = torch.load('/tmp/test.t7')
+print('Gradient estimate error ' .. checkGrad(otherModel, nn.MSECriterion(), input, output))