1 files changed, 14 insertions, 14 deletions
diff --git a/layer/mse.lua b/layer/mse.lua
index da5b24d..9a97add 100644
--- a/layer/mse.lua
+++ b/layer/mse.lua
@@ -8,12 +8,16 @@ function MSELayer:__init(id, global_conf, layer_conf)
     self:check_dim_len(2, -1)
 end
 
-function MSELayer:init()
+function MSELayer:init(batch_size)
     if self.dim_in[1] ~= self.dim_in[2] then
         nerv.error("mismatching dimensions of previous network output and labels")
     end
+    self.scale = 1 / self.dim_in[1]
     self.total_mse = 0.0
     self.total_frames = 0
+    self.mse = self.gconf.cumat_type(batch_size, self.dim_in[1])
+    self.mse_sum = self.gconf.cumat_type(batch_size, 1)
+    self.diff = self.mse:create()
 end
 
 function MSELayer:update(bp_err, input, output)
@@ -21,32 +25,28 @@ function MSELayer:update(bp_err, input, output)
 end
 
 function MSELayer:propagate(input, output)
-    local mse = input[1]:create()
+    local mse = self.mse
+    local mse_sum = self.mse_sum
     mse:add(input[1], input[2], 1.0, -1.0)
-    self.diff = mse:create()
     self.diff:copy_fromd(mse)
     mse:mul_elem(mse, mse)
-    mse = mse:rowsum(mse)
-    local scale = nerv.CuMatrixFloat(mse:nrow(), 1)
-    scale:fill(1 / input[1]:ncol())
-    mse:scale_rows_by_col(scale)
+    mse_sum:add(mse_sum, mse:rowsum(mse), 0.0, self.scale)
     if output[1] ~= nil then
-        output[1]:copy_fromd(mse)
+        output[1]:copy_fromd(mse_sum)
     end
-    self.total_mse = self.total_mse + mse:colsum()[0]
-    self.total_frames = self.total_frames + mse:nrow()
+    self.total_mse = self.total_mse + mse_sum:colsum()[0]
+    self.total_frames = self.total_frames + mse_sum:nrow()
 end
 
 -- NOTE: must call propagate before back_propagate
-function MSELayer:back_propagate(next_bp_err, bp_err, input, output)
+function MSELayer:back_propagate(bp_err, next_bp_err, input, output)
     local nbe = next_bp_err[1]
-    nbe:copy_fromd(self.diff)
-    self.diff = nil
+    nbe:add(nbe, self.diff, 0.0, 2 * self.scale)
     if bp_err[1] ~= nil then
         nbe:scale_rows_by_col(bp_err[1])
     end
 end
 
 function MSELayer:get_params()
-    return {}
+    return nerv.ParamRepo({})
 end