Execution Time0.14s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: master-x86_64-centos7-gcc62-opt-no-rt-cxxmodules (olsnba08.cern.ch) on 2019-11-14 13:00:26
Repository revision: 14de58de35eff907054671888ccc2de0f7f27e77

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 var00.0646768
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      0.313       2.195 
   1 |    -0.5719      0.8251 
   2 |     0.8967      0.8454 
   3 |      -1.37       2.057 
   4 |    0.03905       2.352 
   5 |     0.2068       4.953 
   6 |    -0.9064       1.346 
   7 |      1.437      -4.301 
   8 |     0.9183       2.368 
   9 |    -0.3154      0.1193 

output BN 
output DL feature 0 mean 0.0646768	output DL std 0.878015
output DL feature 1 mean 1.27595	output DL std 2.36075
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.2981      0.4103 
   1 |    -0.7642     -0.2013 
   2 |     0.9988     -0.1922 
   3 |     -1.723      0.3487 
   4 |   -0.03076      0.4802 
   5 |     0.1707       1.642 
   6 |     -1.166     0.03141 
   7 |      1.647       -2.49 
   8 |      1.025      0.4878 
   9 |    -0.4562     -0.5164 

output BN feature 0 mean -2.77556e-17	output BN std 1.05402
output BN feature 1 mean 6.66134e-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 |    0.04309    -0.04111      0.0378    0.003526 
   1 |    0.01037      0.0066     0.02028   0.0002231 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |    0.07108      0.6767      0.6768      -0.236 
   1 |      1.166      -1.947      0.2473      0.3122 

 training batch 2 mu var00.0646796
compute loss for weight  0.0710859  0.0710759 result 0.15778
 training batch 3 mu var00.0646768
compute loss for weight  0.0710659  0.0710759 result 0.157779
 training batch 4 mu var00.0646775
compute loss for weight  0.0710809  0.0710759 result 0.15778
 training batch 5 mu var00.0646768
compute loss for weight  0.0710709  0.0710759 result 0.15778
   --dy = 0.0430907 dy_ref = 0.0430907
 training batch 6 mu var00.0646763
compute loss for weight  0.676664  0.676654 result 0.157779
 training batch 7 mu var00.0646768
compute loss for weight  0.676644  0.676654 result 0.15778
 training batch 8 mu var00.0646766
compute loss for weight  0.676659  0.676654 result 0.15778
 training batch 9 mu var00.0646768
compute loss for weight  0.676649  0.676654 result 0.15778
   --dy = -0.0411067 dy_ref = -0.0411067
 training batch 10 mu var00.0646771
compute loss for weight  0.676814  0.676804 result 0.15778
 training batch 11 mu var00.0646768
compute loss for weight  0.676794  0.676804 result 0.157779
 training batch 12 mu var00.0646769
compute loss for weight  0.676809  0.676804 result 0.15778
 training batch 13 mu var00.0646768
compute loss for weight  0.676799  0.676804 result 0.15778
   --dy = 0.0377993 dy_ref = 0.0377993
 training batch 14 mu var00.0646767
compute loss for weight  -0.236008  -0.236018 result 0.15778
 training batch 15 mu var00.0646768
compute loss for weight  -0.236028  -0.236018 result 0.15778
 training batch 16 mu var00.0646767
compute loss for weight  -0.236013  -0.236018 result 0.15778
 training batch 17 mu var00.0646768
compute loss for weight  -0.236023  -0.236018 result 0.15778
   --dy = 0.00352574 dy_ref = 0.00352574
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |   -0.01163      0.3272 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var00.0646768
compute loss for weight  1.00001  1 result 0.15778
 training batch 19 mu var00.0646768
compute loss for weight  0.99999  1 result 0.15778
 training batch 20 mu var00.0646768
compute loss for weight  1.00001  1 result 0.15778
 training batch 21 mu var00.0646768
compute loss for weight  0.999995  1 result 0.15778
   --dy = -0.0116267 dy_ref = -0.0116267
 training batch 22 mu var00.0646768
compute loss for weight  1.00001  1 result 0.157783
 training batch 23 mu var00.0646768
compute loss for weight  0.99999  1 result 0.157777
 training batch 24 mu var00.0646768
compute loss for weight  1.00001  1 result 0.157781
 training batch 25 mu var00.0646768
compute loss for weight  0.999995  1 result 0.157778
   --dy = 0.327186 dy_ref = 0.327186
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 | -4.337e-19   1.735e-17 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var00.0646768
compute loss for weight  1e-05  0 result 0.15778
 training batch 27 mu var00.0646768
compute loss for weight  -1e-05  0 result 0.15778
 training batch 28 mu var00.0646768
compute loss for weight  5e-06  0 result 0.15778
 training batch 29 mu var00.0646768
compute loss for weight  -5e-06  0 result 0.15778
   --dy = 3.70074e-12 dy_ref = -4.33681e-19
 training batch 30 mu var00.0646768
compute loss for weight  1e-05  0 result 0.15778
 training batch 31 mu var00.0646768
compute loss for weight  -1e-05  0 result 0.15778
 training batch 32 mu var00.0646768
compute loss for weight  5e-06  0 result 0.15778
 training batch 33 mu var00.0646768
compute loss for weight  -5e-06  0 result 0.15778
   --dy = 0 dy_ref = 1.73472e-17
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.1609     -0.7829 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |   -0.07227     -0.4179 

 training batch 34 mu var00.0646768
compute loss for weight  -0.0722598  -0.0722698 result 0.157781
 training batch 35 mu var00.0646768
compute loss for weight  -0.0722798  -0.0722698 result 0.157778
 training batch 36 mu var00.0646768
compute loss for weight  -0.0722648  -0.0722698 result 0.157781
 training batch 37 mu var00.0646768
compute loss for weight  -0.0722748  -0.0722698 result 0.157779
   --dy = 0.160879 dy_ref = 0.160879
 training batch 38 mu var00.0646768
compute loss for weight  -0.41788  -0.41789 result 0.157772
 training batch 39 mu var00.0646768
compute loss for weight  -0.4179  -0.41789 result 0.157788
 training batch 40 mu var00.0646768
compute loss for weight  -0.417885  -0.41789 result 0.157776
 training batch 41 mu var00.0646768
compute loss for weight  -0.417895  -0.41789 result 0.157784
   --dy = -0.782948 dy_ref = -0.782948
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m7.43776e-10[NON-XML-CHAR-0x1B][39m