Execution Time1.22s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: PR-4279-x86_64-centos7-gcc48-opt (olhswep22.cern.ch) on 2019-11-14 21:08:07

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 var00.66968
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      1.061      0.2189 
   1 |     0.4138      -1.557 
   2 |      1.893       1.245 
   3 |      2.253      -1.037 
   4 |      1.483      0.1651 
   5 |      2.764    -0.03093 
   6 |     -2.429      0.7391 
   7 |    -0.4757     -0.6833 
   8 |    -0.3582      0.8383 
   9 |    0.09224     -0.1746 

output BN 
output DL feature 0 mean 0.66968	output DL std 1.54837
output DL feature 1 mean -0.0275645	output DL std 0.873268
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.2662      0.2974 
   1 |    -0.1742      -1.846 
   2 |     0.8329       1.536 
   3 |      1.078      -1.218 
   4 |     0.5536      0.2325 
   5 |      1.426   -0.004062 
   6 |      -2.11      0.9253 
   7 |    -0.7797     -0.7915 
   8 |    -0.6997       1.045 
   9 |    -0.3931     -0.1775 

output BN feature 0 mean 4.44089e-17	output BN std 1.05407
output BN feature 1 mean 4.71845e-17	output BN std 1.05402
Testing weight gradients   for    layer 0
weight gradient for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |   -0.02622     0.03802     -0.1639    -0.05527 
   1 |     -1.436       1.971      -12.77       -6.21 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |     0.6248     -0.4523      0.2544      -1.369 
   1 |    -0.7427     -0.3606       1.042      0.3518 

 training batch 2 mu var00.669683
compute loss for weight  0.624814  0.624804 result 4.30153
 training batch 3 mu var00.66968
compute loss for weight  0.624794  0.624804 result 4.30153
 training batch 4 mu var00.66968
compute loss for weight  0.624809  0.624804 result 4.30153
 training batch 5 mu var00.66968
compute loss for weight  0.624799  0.624804 result 4.30153
   --dy = -0.0262241 dy_ref = -0.0262241
 training batch 6 mu var00.669679
compute loss for weight  -0.452258  -0.452268 result 4.30153
 training batch 7 mu var00.66968
compute loss for weight  -0.452278  -0.452268 result 4.30153
 training batch 8 mu var00.66968
compute loss for weight  -0.452263  -0.452268 result 4.30153
 training batch 9 mu var00.66968
compute loss for weight  -0.452273  -0.452268 result 4.30153
   --dy = 0.0380207 dy_ref = 0.0380207
 training batch 10 mu var00.66968
compute loss for weight  0.254378  0.254368 result 4.30153
 training batch 11 mu var00.66968
compute loss for weight  0.254358  0.254368 result 4.30153
 training batch 12 mu var00.66968
compute loss for weight  0.254373  0.254368 result 4.30153
 training batch 13 mu var00.66968
compute loss for weight  0.254363  0.254368 result 4.30153
   --dy = -0.163902 dy_ref = -0.163902
 training batch 14 mu var00.66968
compute loss for weight  -1.36894  -1.36895 result 4.30153
 training batch 15 mu var00.66968
compute loss for weight  -1.36896  -1.36895 result 4.30153
 training batch 16 mu var00.66968
compute loss for weight  -1.36895  -1.36895 result 4.30153
 training batch 17 mu var00.66968
compute loss for weight  -1.36896  -1.36895 result 4.30153
   --dy = -0.0552742 dy_ref = -0.0552742
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      8.539     0.06359 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var00.66968
compute loss for weight  1.00001  1 result 4.30162
 training batch 19 mu var00.66968
compute loss for weight  0.99999  1 result 4.30145
 training batch 20 mu var00.66968
compute loss for weight  1.00001  1 result 4.30157
 training batch 21 mu var00.66968
compute loss for weight  0.999995  1 result 4.30149
   --dy = 8.53948 dy_ref = 8.53948
 training batch 22 mu var00.66968
compute loss for weight  1.00001  1 result 4.30153
 training batch 23 mu var00.66968
compute loss for weight  0.99999  1 result 4.30153
 training batch 24 mu var00.66968
compute loss for weight  1.00001  1 result 4.30153
 training batch 25 mu var00.66968
compute loss for weight  0.999995  1 result 4.30153
   --dy = 0.0635862 dy_ref = 0.0635862
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |  7.772e-16  -2.082e-17 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var00.66968
compute loss for weight  1e-05  0 result 4.30153
 training batch 27 mu var00.66968
compute loss for weight  -1e-05  0 result 4.30153
 training batch 28 mu var00.66968
compute loss for weight  5e-06  0 result 4.30153
 training batch 29 mu var00.66968
compute loss for weight  -5e-06  0 result 4.30153
   --dy = -1.18424e-10 dy_ref = 7.77156e-16
 training batch 30 mu var00.66968
compute loss for weight  1e-05  0 result 4.30153
 training batch 31 mu var00.66968
compute loss for weight  -1e-05  0 result 4.30153
 training batch 32 mu var00.66968
compute loss for weight  5e-06  0 result 4.30153
 training batch 33 mu var00.66968
compute loss for weight  -5e-06  0 result 4.30153
   --dy = 1.18424e-10 dy_ref = -2.08167e-17
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      4.145      -0.783 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |       2.06    -0.08121 

 training batch 34 mu var00.66968
compute loss for weight  2.06029  2.06028 result 4.30157
 training batch 35 mu var00.66968
compute loss for weight  2.06027  2.06028 result 4.30149
 training batch 36 mu var00.66968
compute loss for weight  2.06028  2.06028 result 4.30155
 training batch 37 mu var00.66968
compute loss for weight  2.06027  2.06028 result 4.30151
   --dy = 4.14482 dy_ref = 4.14482
 training batch 38 mu var00.66968
compute loss for weight  -0.0811963  -0.0812063 result 4.30152
 training batch 39 mu var00.66968
compute loss for weight  -0.0812163  -0.0812063 result 4.30154
 training batch 40 mu var00.66968
compute loss for weight  -0.0812013  -0.0812063 result 4.30153
 training batch 41 mu var00.66968
compute loss for weight  -0.0812113  -0.0812063 result 4.30154
   --dy = -0.783021 dy_ref = -0.783021
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][33m4.28723e-09[NON-XML-CHAR-0x1B][39m