require 'torch'
require 'nn'
+
+-- require 'cunn'
+
require 'dagnn'
--- torch.setnumthreads(params.nbThreads)
torch.setdefaulttensortype('torch.DoubleTensor')
-torch.manualSeed(2)
+torch.manualSeed(1)
-function checkGrad(model, criterion, input, target)
+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)
local num = (loss1 - loss0) / (2 * epsilon)
if num ~= ana then
- err = math.max(err, torch.abs(num - ana) / torch.abs(num))
+ err = math.max(err, math.abs(num - ana) / math.max(epsilon, math.abs(num)))
end
end
end
end
--- +- Linear(10, 10) -> ReLU ---> d --+
--- / / \
--- / / \
--- --> a --> b -----------> c --------------+ e -->
--- \ /
--- \ /
--- +-- Mul(-1) --------+
+-- +-- Linear(10, 10) --> ReLU --> d -->
+-- / /
+-- / /
+-- --> a --> b -----------> c ---------------+
+-- \
+-- \
+-- +--------------- e -->
-model = nn.DAG()
+dag = nn.DAG()
a = nn.Linear(50, 10)
b = nn.ReLU()
c = nn.Linear(10, 15)
d = nn.CMulTable()
-e = nn.CAddTable()
-
-model:addEdge(a, b)
-model:addEdge(b, nn.Linear(10, 15), nn.ReLU(), d)
-model:addEdge(d, e)
-model:addEdge(b, c)
-model:addEdge(c, d)
-model:addEdge(c, nn.Mul(-1), e)
-
-model:setInput(a)
-model:setOutput(e)
+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
local input = torch.Tensor(30, 50):uniform()
local output = model:updateOutput(input):clone()
-
output:uniform()
-print('Error = ' .. checkGrad(model, nn.MSECriterion(), input, output))
+-- Check that DAG:accGradParameters and friends work okay
+print('Gradient estimate error ' .. checkGrad(model, criterion, input, output, epsilon))
+
+-- 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))
+
+dag:print()