diff --git a/tutorial/cnn.sac b/tutorial/cnn.sac
index c210549..e3243ae 100644
--- a/tutorial/cnn.sac
+++ b/tutorial/cnn.sac
@@ -155,32 +155,29 @@ float[*], float[*] BackConv2( float[*] d_output, float[*] weights, float[*] inpu
 #endif
 //------------------------------------------------------------------------------
 
+
+inline float[*] >--conv-> (float[*] in, float[*] weights) { return Conv (in, weights); }
+inline float[*] >--logistics-> (float[*] in, int _) { return Logistic (in); }
+inline float[*] >--avgpool-> (float[*] in, int[.] filter) { return AveragePool (in, filter, []); }
+inline float[*] >--reshape-> (float[*] in, int[.] s) { return reshape (s, in); }
+
 int main()
 {
    in = genarray( [28,28], 0f);
    in[6,6] = 42f;
-
-   print( in);
-
    k1 = genarray( [5,5,6], 1f/25f);
-   c1 = Logistic( Conv( in, k1 ));
-   print( shape(c1));
-
-   s1 = AveragePool( c1, [2,2], []);
-   print( shape( s1));
-
    k2 = genarray( [5,5,6,12], 1f/150f);
-   c2 = Logistic( Conv( s1, k2));
-
-   print( shape( c2));
-
-   s2 = AveragePool( c2, [2,2], []);
-   print( shape( s2));
-
    fc = genarray( [4,4,1,12,10], 1f/192f);
-   out = reshape( [10], Conv( s2, fc));
-   print( out);
+   _ = 0;
+
+   out = in                                 // Input Layer
+         >--conv-> k1 >--logistics-> _      // Convolution Layer C1
+         >--avgpool-> [2,2]                 // Pooling Layer S1
+         >--conv-> k2 >--logistics-> _      // Convolution Layer C2
+         >--avgpool-> [2,2]                 // Pooling Layer S2
+         >--conv-> fc >--reshape-> [10];    // Fully connection layer FC
 
+   print (out);
    return 0;
 }