Execution Time0.19s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: master-x86_64-fedora29-gcc8-opt-exp-pyroot (root-fedora29-3.cern.ch) on 2019-11-15 04:46:36

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 var0-0.271775
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |       -0.3       0.338 
   1 |    -0.6077     -0.4451 
   2 |    -0.1359       1.365 
   3 |     -1.807      0.3962 
   4 |    -0.6314      0.4865 
   5 |     -1.139      0.7373 
   6 |     0.2977      -1.387 
   7 |       1.24      0.3774 
   8 |     0.6045     -0.1762 
   9 |    -0.2393     -0.0623 

output BN 
output DL feature 0 mean -0.271775	output DL std 0.866092
output DL feature 1 mean 0.163035	output DL std 0.741526
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |   -0.03432      0.2487 
   1 |    -0.4089     -0.8644 
   2 |     0.1653       1.709 
   3 |     -1.869      0.3315 
   4 |    -0.4377      0.4598 
   5 |     -1.055      0.8162 
   6 |     0.6931      -2.203 
   7 |       1.84      0.3047 
   8 |      1.066     -0.4821 
   9 |    0.03947     -0.3203 

output BN feature 0 mean -6.31439e-17	output BN std 1.05401
output BN feature 1 mean -5.55112e-18	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 |    0.09374      0.1577       0.658     -0.8692 
   1 |    -0.8999       1.766      -1.831       4.995 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |    -0.2761      0.6619      0.3508      0.3553 
   1 |   -0.05738     0.03327      0.4974      -0.698 

 training batch 2 mu var0-0.271772
compute loss for weight  -0.276107  -0.276117 result 1.85831
 training batch 3 mu var0-0.271775
compute loss for weight  -0.276127  -0.276117 result 1.85831
 training batch 4 mu var0-0.271774
compute loss for weight  -0.276112  -0.276117 result 1.85831
 training batch 5 mu var0-0.271775
compute loss for weight  -0.276122  -0.276117 result 1.85831
   --dy = 0.0937376 dy_ref = 0.0937376
 training batch 6 mu var0-0.271775
compute loss for weight  0.66191  0.6619 result 1.85831
 training batch 7 mu var0-0.271775
compute loss for weight  0.66189  0.6619 result 1.85831
 training batch 8 mu var0-0.271775
compute loss for weight  0.661905  0.6619 result 1.85831
 training batch 9 mu var0-0.271775
compute loss for weight  0.661895  0.6619 result 1.85831
   --dy = 0.157679 dy_ref = 0.157679
 training batch 10 mu var0-0.271775
compute loss for weight  0.350858  0.350848 result 1.85831
 training batch 11 mu var0-0.271775
compute loss for weight  0.350838  0.350848 result 1.8583
 training batch 12 mu var0-0.271775
compute loss for weight  0.350853  0.350848 result 1.85831
 training batch 13 mu var0-0.271775
compute loss for weight  0.350843  0.350848 result 1.8583
   --dy = 0.657992 dy_ref = 0.657992
 training batch 14 mu var0-0.271775
compute loss for weight  0.355275  0.355265 result 1.8583
 training batch 15 mu var0-0.271775
compute loss for weight  0.355255  0.355265 result 1.85832
 training batch 16 mu var0-0.271775
compute loss for weight  0.35527  0.355265 result 1.8583
 training batch 17 mu var0-0.271775
compute loss for weight  0.35526  0.355265 result 1.85831
   --dy = -0.869201 dy_ref = -0.869201
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      3.671     0.04573 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var0-0.271775
compute loss for weight  1.00001  1 result 1.85834
 training batch 19 mu var0-0.271775
compute loss for weight  0.99999  1 result 1.85827
 training batch 20 mu var0-0.271775
compute loss for weight  1.00001  1 result 1.85833
 training batch 21 mu var0-0.271775
compute loss for weight  0.999995  1 result 1.85829
   --dy = 3.67089 dy_ref = 3.67089
 training batch 22 mu var0-0.271775
compute loss for weight  1.00001  1 result 1.85831
 training batch 23 mu var0-0.271775
compute loss for weight  0.99999  1 result 1.85831
 training batch 24 mu var0-0.271775
compute loss for weight  1.00001  1 result 1.85831
 training batch 25 mu var0-0.271775
compute loss for weight  0.999995  1 result 1.85831
   --dy = 0.0457274 dy_ref = 0.0457274
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |  -2.88e-16  -5.725e-17 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var0-0.271775
compute loss for weight  1e-05  0 result 1.85831
 training batch 27 mu var0-0.271775
compute loss for weight  -1e-05  0 result 1.85831
 training batch 28 mu var0-0.271775
compute loss for weight  5e-06  0 result 1.85831
 training batch 29 mu var0-0.271775
compute loss for weight  -5e-06  0 result 1.85831
   --dy = 0 dy_ref = -2.87964e-16
 training batch 30 mu var0-0.271775
compute loss for weight  1e-05  0 result 1.85831
 training batch 31 mu var0-0.271775
compute loss for weight  -1e-05  0 result 1.85831
 training batch 32 mu var0-0.271775
compute loss for weight  5e-06  0 result 1.85831
 training batch 33 mu var0-0.271775
compute loss for weight  -5e-06  0 result 1.85831
   --dy = 6.66134e-11 dy_ref = -5.72459e-17
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      2.606      0.1104 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      1.409      0.4141 

 training batch 34 mu var0-0.271775
compute loss for weight  1.40867  1.40866 result 1.85833
 training batch 35 mu var0-0.271775
compute loss for weight  1.40865  1.40866 result 1.85828
 training batch 36 mu var0-0.271775
compute loss for weight  1.40867  1.40866 result 1.85832
 training batch 37 mu var0-0.271775
compute loss for weight  1.40866  1.40866 result 1.85829
   --dy = 2.60594 dy_ref = 2.60594
 training batch 38 mu var0-0.271775
compute loss for weight  0.41413  0.41412 result 1.85831
 training batch 39 mu var0-0.271775
compute loss for weight  0.41411  0.41412 result 1.85831
 training batch 40 mu var0-0.271775
compute loss for weight  0.414125  0.41412 result 1.85831
 training batch 41 mu var0-0.271775
compute loss for weight  0.414115  0.41412 result 1.85831
   --dy = 0.110421 dy_ref = 0.110421
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][33m1.12409e-09[NON-XML-CHAR-0x1B][39m