Execution Time0.18s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: master-x86_64-centos7-gcc48-dbg (olhswep22.cern.ch) on 2019-11-13 14:19:10

Test Timing: Passed
Processors1

Show Command Line
Display graphs:

Test output
Testing Backpropagation:
DEEP NEURAL NETWORK:   Depth = 3  Input = ( 1, 10, 4 )  Batch size = 10  Loss function = R
	Layer 0	 DENSE Layer: 	 ( Input =     4 , Width =     2 ) 	Output = (  1 ,    10 ,     2 ) 	 Activation Function = Identity
	Layer 1	 BATCH NORM Layer: 	 ( Input =     2 ) 
	Layer 2	 DENSE Layer: 	 ( Input =     2 , Width =     1 ) 	Output = (  1 ,    10 ,     1 ) 	 Activation Function = Identity
input 

10x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |     0.9989     -0.4348      0.7818    -0.03005 
   1 |     0.8243    -0.05672     -0.9009     -0.0747 
   2 |   0.007912     -0.4108       1.391     -0.9851 
   3 |   -0.04894      -1.443      -1.061      -1.388 
   4 |     0.7674      -0.736      0.5797     -0.3821 
   5 |      2.061      -1.235       1.165     -0.4542 
   6 |    -0.1348     -0.4996     -0.1824       1.844 
   7 |    -0.2428       1.997    0.004806     -0.4222 
   8 |      1.541     0.09474       1.525       1.217 
   9 |    -0.1363     -0.1992     -0.2938     -0.1184 

 training batch 1 mu var00.339162
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.4321      0.6052 
   1 |       1.13     0.01819 
   2 |    -0.8216      0.6981 
   3 |      1.041      0.9494 
   4 |     0.4716      0.7796 
   5 |       1.25       1.446 
   6 |     0.6158    -0.05025 
   7 |     -1.229       -1.25 
   8 |     0.3444      0.2897 
   9 |     0.1563     0.06814 

output BN 
output DL feature 0 mean 0.339162	output DL std 0.809043
output DL feature 1 mean 0.355434	output DL std 0.733629
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.1211      0.3589 
   1 |      1.031     -0.4845 
   2 |     -1.512      0.4923 
   3 |     0.9148      0.8533 
   4 |     0.1725      0.6094 
   5 |      1.187       1.567 
   6 |     0.3604     -0.5828 
   7 |     -2.043      -2.306 
   8 |   0.006808    -0.09447 
   9 |    -0.2382     -0.4127 

output BN feature 0 mean 1.66533e-17	output BN std 1.054
output BN feature 1 mean 0	output BN std 1.05398
Testing weight gradients   for    layer 0
weight gradient for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |      1.278      -3.062       3.145      -2.547 
   1 |     -2.932        6.75      -6.066       4.995 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |      0.697     -0.4883     -0.6021       0.193 
   1 |     0.1718     -0.6413      0.1915     -0.1695 

 training batch 2 mu var00.339165
compute loss for weight  0.697034  0.697024 result 10.8596
 training batch 3 mu var00.339162
compute loss for weight  0.697014  0.697024 result 10.8595
 training batch 4 mu var00.339163
compute loss for weight  0.697029  0.697024 result 10.8596
 training batch 5 mu var00.339162
compute loss for weight  0.697019  0.697024 result 10.8595
   --dy = 1.27806 dy_ref = 1.27806
 training batch 6 mu var00.339161
compute loss for weight  -0.488333  -0.488343 result 10.8595
 training batch 7 mu var00.339162
compute loss for weight  -0.488353  -0.488343 result 10.8596
 training batch 8 mu var00.339162
compute loss for weight  -0.488338  -0.488343 result 10.8595
 training batch 9 mu var00.339162
compute loss for weight  -0.488348  -0.488343 result 10.8596
   --dy = -3.0622 dy_ref = -3.0622
 training batch 10 mu var00.339162
compute loss for weight  -0.602053  -0.602063 result 10.8596
 training batch 11 mu var00.339162
compute loss for weight  -0.602073  -0.602063 result 10.8595
 training batch 12 mu var00.339162
compute loss for weight  -0.602058  -0.602063 result 10.8596
 training batch 13 mu var00.339162
compute loss for weight  -0.602068  -0.602063 result 10.8595
   --dy = 3.1451 dy_ref = 3.1451
 training batch 14 mu var00.339162
compute loss for weight  0.193005  0.192995 result 10.8595
 training batch 15 mu var00.339162
compute loss for weight  0.192985  0.192995 result 10.8596
 training batch 16 mu var00.339162
compute loss for weight  0.193  0.192995 result 10.8595
 training batch 17 mu var00.339162
compute loss for weight  0.19299  0.192995 result 10.8596
   --dy = -2.54745 dy_ref = -2.54745
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      6.207       15.51 

weights for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |          1           1 

 training batch 18 mu var00.339162
compute loss for weight  1.00001  1 result 10.8596
 training batch 19 mu var00.339162
compute loss for weight  0.99999  1 result 10.8595
 training batch 20 mu var00.339162
compute loss for weight  1.00001  1 result 10.8596
 training batch 21 mu var00.339162
compute loss for weight  0.999995  1 result 10.8595
   --dy = 6.20675 dy_ref = 6.20675
 training batch 22 mu var00.339162
compute loss for weight  1.00001  1 result 10.8597
 training batch 23 mu var00.339162
compute loss for weight  0.99999  1 result 10.8594
 training batch 24 mu var00.339162
compute loss for weight  1.00001  1 result 10.8596
 training batch 25 mu var00.339162
compute loss for weight  0.999995  1 result 10.8595
   --dy = 15.5123 dy_ref = 15.5123
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 | -1.665e-16  -3.331e-16 

weights for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |          0           0 

 training batch 26 mu var00.339162
compute loss for weight  1e-05  0 result 10.8595
 training batch 27 mu var00.339162
compute loss for weight  -1e-05  0 result 10.8595
 training batch 28 mu var00.339162
compute loss for weight  5e-06  0 result 10.8595
 training batch 29 mu var00.339162
compute loss for weight  -5e-06  0 result 10.8595
   --dy = 4.14483e-10 dy_ref = -1.66533e-16
 training batch 30 mu var00.339162
compute loss for weight  1e-05  0 result 10.8595
 training batch 31 mu var00.339162
compute loss for weight  -1e-05  0 result 10.8595
 training batch 32 mu var00.339162
compute loss for weight  5e-06  0 result 10.8595
 training batch 33 mu var00.339162
compute loss for weight  -5e-06  0 result 10.8595
   --dy = 5.32907e-10 dy_ref = -3.33067e-16
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      5.337        6.33 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      1.163       2.451 

 training batch 34 mu var00.339162
compute loss for weight  1.16299  1.16298 result 10.8596
 training batch 35 mu var00.339162
compute loss for weight  1.16297  1.16298 result 10.8595
 training batch 36 mu var00.339162
compute loss for weight  1.16298  1.16298 result 10.8596
 training batch 37 mu var00.339162
compute loss for weight  1.16297  1.16298 result 10.8595
   --dy = 5.33695 dy_ref = 5.33695
 training batch 38 mu var00.339162
compute loss for weight  2.45079  2.45078 result 10.8596
 training batch 39 mu var00.339162
compute loss for weight  2.45077  2.45078 result 10.8595
 training batch 40 mu var00.339162
compute loss for weight  2.45078  2.45078 result 10.8596
 training batch 41 mu var00.339162
compute loss for weight  2.45077  2.45078 result 10.8595
   --dy = 6.32955 dy_ref = 6.32955
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m5.32907e-10[NON-XML-CHAR-0x1B][39m