CMulTable.lua

require 'nn'

-- monkey patch it in :-P

nn.CMulTable.baseUpdateOutput = nn.CMulTable.updateOutput
nn.CMulTable.baseUpdateGradInput = nn.CMulTable.updateGradInput

function nn.CMulTable:updateOutput(input, target)
   if torch.type(input[1]) ~= 'torch.ClTensor' then
      return self:baseUpdateOutput(input, target)
   end
   -- copied from nn for now
   self.output:resizeAs(input[1])
   if true then
      if #input == 2 then  -- this is really common case
         -- anyway, it's what char-rnn uses :-P
         self.output:cmul(input[1], input[2])
      else
         self.output:copy(input[1])
         for i=2,#input do
            self.output:cmul(input[i])
         end
      end
   end
   return self.output
end


function nn.CMulTable:updateGradInput(input, gradOutput)
   --  print('cl.updategradinput', torch.type(input), torch.type(gradOutput))
   if torch.type(gradOutput) ~= 'torch.ClTensor' then
      return self:baseUpdateGradInput(input, gradOutput)
   end
   -- copied from nn for now
   for i=1,#input do
      self.gradInput[i] = self.gradInput[i] or input[1].new()
      self.gradInput[i]:resizeAs(input[i])
      if true then
         if #input == 2 then
            self.gradInput[i]:cmul(gradOutput, input[2-i + 1])
         else
            self.gradInput[i]:copy(gradOutput)
            for j=1,#input do
               if i~=j then
                  self.gradInput[i]:cmul(input[j])
               end
            end
         end
      end
   end
   return self.gradInput
end