Execution Time0.05s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: PR-4279-x86_64-ubuntu16-gcc54-opt (sft-ubuntu-1604-4) on 2019-11-14 21:01:42
Repository revision: eb9d2d64c365eec560379f62009bcc1579861643

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 var0-0.0538178
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.4226       -2.84 
   1 |    -0.1194      0.5456 
   2 |      1.295      -3.216 
   3 |      3.918      0.7208 
   4 |     0.6279      -2.481 
   5 |      0.192      -5.378 
   6 |      -1.65        1.32 
   7 |     -1.652      0.7792 
   8 |     -3.315      -4.002 
   9 |     0.5878      0.5687 

output BN 
output DL feature 0 mean -0.0538178	output DL std 1.95517
output DL feature 1 mean -1.39834	output DL std 2.43674
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.1988     -0.6234 
   1 |   -0.03537      0.8409 
   2 |      0.727     -0.7865 
   3 |      2.141      0.9167 
   4 |     0.3675     -0.4685 
   5 |     0.1325      -1.722 
   6 |    -0.8606       1.176 
   7 |    -0.8615       0.942 
   8 |     -1.758      -1.126 
   9 |     0.3459      0.8509 

output BN feature 0 mean -2.22045e-17	output BN std 1.05408
output BN feature 1 mean -5.55112e-17	output BN std 1.05408
Testing weight gradients   for    layer 0
weight gradient for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |      1.335     -0.8114       1.661     -0.4237 
   1 |      1.663      0.4177       2.265         1.2 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |    -0.7147      -1.189     -0.3385      -1.302 
   1 |     -1.249      0.3571      -1.816      0.5412 

 training batch 2 mu var0-0.053815
compute loss for weight  -0.714677  -0.714687 result 4.2718
 training batch 3 mu var0-0.0538178
compute loss for weight  -0.714697  -0.714687 result 4.27178
 training batch 4 mu var0-0.0538171
compute loss for weight  -0.714682  -0.714687 result 4.2718
 training batch 5 mu var0-0.0538178
compute loss for weight  -0.714692  -0.714687 result 4.27178
   --dy = 1.33451 dy_ref = 1.33451
 training batch 6 mu var0-0.0538183
compute loss for weight  -1.18866  -1.18867 result 4.27178
 training batch 7 mu var0-0.0538178
compute loss for weight  -1.18868  -1.18867 result 4.2718
 training batch 8 mu var0-0.053818
compute loss for weight  -1.18866  -1.18867 result 4.27179
 training batch 9 mu var0-0.0538178
compute loss for weight  -1.18867  -1.18867 result 4.27179
   --dy = -0.811424 dy_ref = -0.811424
 training batch 10 mu var0-0.0538175
compute loss for weight  -0.338483  -0.338493 result 4.27181
 training batch 11 mu var0-0.0538178
compute loss for weight  -0.338503  -0.338493 result 4.27177
 training batch 12 mu var0-0.0538177
compute loss for weight  -0.338488  -0.338493 result 4.2718
 training batch 13 mu var0-0.0538178
compute loss for weight  -0.338498  -0.338493 result 4.27178
   --dy = 1.66147 dy_ref = 1.66147
 training batch 14 mu var0-0.0538179
compute loss for weight  -1.30246  -1.30247 result 4.27179
 training batch 15 mu var0-0.0538178
compute loss for weight  -1.30248  -1.30247 result 4.27179
 training batch 16 mu var0-0.0538178
compute loss for weight  -1.30247  -1.30247 result 4.27179
 training batch 17 mu var0-0.0538178
compute loss for weight  -1.30248  -1.30247 result 4.27179
   --dy = -0.423683 dy_ref = -0.423683
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      6.722       1.822 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var0-0.0538178
compute loss for weight  1.00001  1 result 4.27186
 training batch 19 mu var0-0.0538178
compute loss for weight  0.99999  1 result 4.27172
 training batch 20 mu var0-0.0538178
compute loss for weight  1.00001  1 result 4.27182
 training batch 21 mu var0-0.0538178
compute loss for weight  0.999995  1 result 4.27176
   --dy = 6.72165 dy_ref = 6.72165
 training batch 22 mu var0-0.0538178
compute loss for weight  1.00001  1 result 4.27181
 training batch 23 mu var0-0.0538178
compute loss for weight  0.99999  1 result 4.27177
 training batch 24 mu var0-0.0538178
compute loss for weight  1.00001  1 result 4.2718
 training batch 25 mu var0-0.0538178
compute loss for weight  0.999995  1 result 4.27178
   --dy = 1.82193 dy_ref = 1.82193
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 | -3.331e-16   -2.22e-16 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var0-0.0538178
compute loss for weight  1e-05  0 result 4.27179
 training batch 27 mu var0-0.0538178
compute loss for weight  -1e-05  0 result 4.27179
 training batch 28 mu var0-0.0538178
compute loss for weight  5e-06  0 result 4.27179
 training batch 29 mu var0-0.0538178
compute loss for weight  -5e-06  0 result 4.27179
   --dy = 1.18424e-10 dy_ref = -3.33067e-16
 training batch 30 mu var0-0.0538178
compute loss for weight  1e-05  0 result 4.27179
 training batch 31 mu var0-0.0538178
compute loss for weight  -1e-05  0 result 4.27179
 training batch 32 mu var0-0.0538178
compute loss for weight  5e-06  0 result 4.27179
 training batch 33 mu var0-0.0538178
compute loss for weight  -5e-06  0 result 4.27179
   --dy = -1.4803e-11 dy_ref = -2.22045e-16
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     -3.499       1.637 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     -1.921       1.113 

 training batch 34 mu var0-0.0538178
compute loss for weight  -1.92074  -1.92075 result 4.27176
 training batch 35 mu var0-0.0538178
compute loss for weight  -1.92076  -1.92075 result 4.27183
 training batch 36 mu var0-0.0538178
compute loss for weight  -1.92074  -1.92075 result 4.27177
 training batch 37 mu var0-0.0538178
compute loss for weight  -1.92075  -1.92075 result 4.27181
   --dy = -3.4995 dy_ref = -3.4995
 training batch 38 mu var0-0.0538178
compute loss for weight  1.11326  1.11325 result 4.27181
 training batch 39 mu var0-0.0538178
compute loss for weight  1.11324  1.11325 result 4.27177
 training batch 40 mu var0-0.0538178
compute loss for weight  1.11325  1.11325 result 4.2718
 training batch 41 mu var0-0.0538178
compute loss for weight  1.11324  1.11325 result 4.27178
   --dy = 1.63659 dy_ref = 1.63659
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m3.96673e-10[NON-XML-CHAR-0x1B][39m