Execution Time0.62s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: PR-420-x86_64-fedora29-gcc8-opt (root-fedora29-2.cern.ch) on 2019-11-14 09:35:44

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 var00.182118
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.4435      0.3686 
   1 |     0.6299     -0.7281 
   2 |    -0.6005       1.022 
   3 |      -1.92      -1.158 
   4 |    -0.1674      0.1617 
   5 |      0.455      0.3493 
   6 |    -0.2645     -0.5907 
   7 |       1.83       0.859 
   8 |      1.784      0.9255 
   9 |    -0.3694     -0.2741 

output BN 
output DL feature 0 mean 0.182118	output DL std 1.11979
output DL feature 1 mean 0.0935646	output DL std 0.754984
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      0.246      0.3839 
   1 |     0.4215      -1.147 
   2 |    -0.7367       1.296 
   3 |     -1.979      -1.747 
   4 |     -0.329     0.09508 
   5 |     0.2569       0.357 
   6 |    -0.4204     -0.9553 
   7 |      1.551       1.069 
   8 |      1.508       1.161 
   9 |    -0.5192     -0.5132 

output BN feature 0 mean -3.33067e-17	output BN std 1.05405
output BN feature 1 mean -4.44089e-17	output BN std 1.05399
Testing weight gradients   for    layer 0
weight gradient for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |  -0.002024   -0.001435     -0.0646     0.02618 
   1 |     -1.256      -2.553      -4.452     -0.2351 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |     0.8917        1.07     0.03143      0.2148 
   1 |   -0.05096      0.3917      0.7489     -0.1438 

 training batch 2 mu var00.182121
compute loss for weight  0.891719  0.891709 result 1.39114
 training batch 3 mu var00.182118
compute loss for weight  0.891699  0.891709 result 1.39114
 training batch 4 mu var00.182119
compute loss for weight  0.891714  0.891709 result 1.39114
 training batch 5 mu var00.182118
compute loss for weight  0.891704  0.891709 result 1.39114
   --dy = -0.00202383 dy_ref = -0.00202383
 training batch 6 mu var00.182118
compute loss for weight  1.0704  1.07039 result 1.39114
 training batch 7 mu var00.182118
compute loss for weight  1.07038  1.07039 result 1.39114
 training batch 8 mu var00.182118
compute loss for weight  1.07039  1.07039 result 1.39114
 training batch 9 mu var00.182118
compute loss for weight  1.07038  1.07039 result 1.39114
   --dy = -0.00143542 dy_ref = -0.00143542
 training batch 10 mu var00.182119
compute loss for weight  0.03144  0.03143 result 1.39114
 training batch 11 mu var00.182118
compute loss for weight  0.03142  0.03143 result 1.39114
 training batch 12 mu var00.182119
compute loss for weight  0.031435  0.03143 result 1.39114
 training batch 13 mu var00.182118
compute loss for weight  0.031425  0.03143 result 1.39114
   --dy = -0.0646037 dy_ref = -0.0646037
 training batch 14 mu var00.182118
compute loss for weight  0.214855  0.214845 result 1.39114
 training batch 15 mu var00.182118
compute loss for weight  0.214835  0.214845 result 1.39114
 training batch 16 mu var00.182118
compute loss for weight  0.21485  0.214845 result 1.39114
 training batch 17 mu var00.182118
compute loss for weight  0.21484  0.214845 result 1.39114
   --dy = 0.0261804 dy_ref = 0.0261804
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      2.857    -0.07461 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var00.182118
compute loss for weight  1.00001  1 result 1.39117
 training batch 19 mu var00.182118
compute loss for weight  0.99999  1 result 1.39111
 training batch 20 mu var00.182118
compute loss for weight  1.00001  1 result 1.39116
 training batch 21 mu var00.182118
compute loss for weight  0.999995  1 result 1.39113
   --dy = 2.85689 dy_ref = 2.85689
 training batch 22 mu var00.182118
compute loss for weight  1.00001  1 result 1.39114
 training batch 23 mu var00.182118
compute loss for weight  0.99999  1 result 1.39114
 training batch 24 mu var00.182118
compute loss for weight  1.00001  1 result 1.39114
 training batch 25 mu var00.182118
compute loss for weight  0.999995  1 result 1.39114
   --dy = -0.0746114 dy_ref = -0.0746114
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 | -2.776e-17  -3.469e-18 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var00.182118
compute loss for weight  1e-05  0 result 1.39114
 training batch 27 mu var00.182118
compute loss for weight  -1e-05  0 result 1.39114
 training batch 28 mu var00.182118
compute loss for weight  5e-06  0 result 1.39114
 training batch 29 mu var00.182118
compute loss for weight  -5e-06  0 result 1.39114
   --dy = -3.70074e-12 dy_ref = -2.77556e-17
 training batch 30 mu var00.182118
compute loss for weight  1e-05  0 result 1.39114
 training batch 31 mu var00.182118
compute loss for weight  -1e-05  0 result 1.39114
 training batch 32 mu var00.182118
compute loss for weight  5e-06  0 result 1.39114
 training batch 33 mu var00.182118
compute loss for weight  -5e-06  0 result 1.39114
   --dy = 0 dy_ref = -3.46945e-18
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     -2.357      -1.409 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     -1.212     0.05295 

 training batch 34 mu var00.182118
compute loss for weight  -1.21188  -1.21189 result 1.39112
 training batch 35 mu var00.182118
compute loss for weight  -1.2119  -1.21189 result 1.39116
 training batch 36 mu var00.182118
compute loss for weight  -1.21189  -1.21189 result 1.39113
 training batch 37 mu var00.182118
compute loss for weight  -1.2119  -1.21189 result 1.39115
   --dy = -2.35738 dy_ref = -2.35738
 training batch 38 mu var00.182118
compute loss for weight  0.0529629  0.0529529 result 1.39113
 training batch 39 mu var00.182118
compute loss for weight  0.0529429  0.0529529 result 1.39116
 training batch 40 mu var00.182118
compute loss for weight  0.0529579  0.0529529 result 1.39113
 training batch 41 mu var00.182118
compute loss for weight  0.0529479  0.0529529 result 1.39115
   --dy = -1.40902 dy_ref = -1.40902
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][33m2.11841e-08[NON-XML-CHAR-0x1B][39m