Execution Time0.15s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: master-x86_64-centos7-clang100-opt-no-rt-cxxmodules (olsnba08.cern.ch) on 2019-11-16 01:37:25
Repository revision: cdf8874e8c83beaf9bc39a224629667fbb7903bc

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 var0-0.0512786
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.1243     -0.9471 
   1 |     0.2761     -0.4356 
   2 |    -0.9943     0.09728 
   3 |    -0.1315     -0.3439 
   4 |    -0.2036     -0.8601 
   5 |     -0.267      -1.996 
   6 |      1.468       -1.24 
   7 |    -0.9291       1.923 
   8 |      0.315      -1.517 
   9 |    0.07812   -0.001607 

output BN 
output DL feature 0 mean -0.0512786	output DL std 0.691764
output DL feature 1 mean -0.532158	output DL std 1.0874
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.1112     -0.4022 
   1 |     0.4988     0.09356 
   2 |     -1.437      0.6101 
   3 |    -0.1223      0.1825 
   4 |     -0.232     -0.3179 
   5 |    -0.3286      -1.419 
   6 |      2.314     -0.6857 
   7 |     -1.337       2.379 
   8 |     0.5581     -0.9549 
   9 |     0.1972      0.5143 

output BN feature 0 mean -2.22045e-17	output BN std 1.05397
output BN feature 1 mean -1.11022e-17	output BN std 1.05404
Testing weight gradients   for    layer 0
weight gradient for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |     -1.045       1.013      -0.867     0.07949 
   1 |     0.3648     -0.6938     -0.1349       1.115 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |     0.0146     -0.3194     -0.3292      0.6778 
   1 |    -0.5794      0.7852    -0.05393      -0.507 

 training batch 2 mu var0-0.0512757
compute loss for weight  0.0146141  0.0146041 result 0.978783
 training batch 3 mu var0-0.0512786
compute loss for weight  0.0145941  0.0146041 result 0.978804
 training batch 4 mu var0-0.0512779
compute loss for weight  0.0146091  0.0146041 result 0.978788
 training batch 5 mu var0-0.0512786
compute loss for weight  0.0145991  0.0146041 result 0.978799
   --dy = -1.04531 dy_ref = -1.04531
 training batch 6 mu var0-0.051279
compute loss for weight  -0.319386  -0.319396 result 0.978804
 training batch 7 mu var0-0.0512786
compute loss for weight  -0.319406  -0.319396 result 0.978784
 training batch 8 mu var0-0.0512787
compute loss for weight  -0.319391  -0.319396 result 0.978799
 training batch 9 mu var0-0.0512786
compute loss for weight  -0.319401  -0.319396 result 0.978789
   --dy = 1.01298 dy_ref = 1.01298
 training batch 10 mu var0-0.0512783
compute loss for weight  -0.329182  -0.329192 result 0.978785
 training batch 11 mu var0-0.0512786
compute loss for weight  -0.329202  -0.329192 result 0.978802
 training batch 12 mu var0-0.0512784
compute loss for weight  -0.329187  -0.329192 result 0.978789
 training batch 13 mu var0-0.0512786
compute loss for weight  -0.329197  -0.329192 result 0.978798
   --dy = -0.867 dy_ref = -0.867
 training batch 14 mu var0-0.0512786
compute loss for weight  0.677804  0.677794 result 0.978794
 training batch 15 mu var0-0.0512786
compute loss for weight  0.677784  0.677794 result 0.978793
 training batch 16 mu var0-0.0512786
compute loss for weight  0.677799  0.677794 result 0.978794
 training batch 17 mu var0-0.0512786
compute loss for weight  0.677789  0.677794 result 0.978793
   --dy = 0.0794912 dy_ref = 0.0794912
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      2.066     -0.1082 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var0-0.0512786
compute loss for weight  1.00001  1 result 0.978814
 training batch 19 mu var0-0.0512786
compute loss for weight  0.99999  1 result 0.978773
 training batch 20 mu var0-0.0512786
compute loss for weight  1.00001  1 result 0.978804
 training batch 21 mu var0-0.0512786
compute loss for weight  0.999995  1 result 0.978783
   --dy = 2.06579 dy_ref = 2.06579
 training batch 22 mu var0-0.0512786
compute loss for weight  1.00001  1 result 0.978793
 training batch 23 mu var0-0.0512786
compute loss for weight  0.99999  1 result 0.978795
 training batch 24 mu var0-0.0512786
compute loss for weight  1.00001  1 result 0.978793
 training batch 25 mu var0-0.0512786
compute loss for weight  0.999995  1 result 0.978794
   --dy = -0.108199 dy_ref = -0.108199
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |  5.551e-17  -1.388e-17 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var0-0.0512786
compute loss for weight  1e-05  0 result 0.978794
 training batch 27 mu var0-0.0512786
compute loss for weight  -1e-05  0 result 0.978794
 training batch 28 mu var0-0.0512786
compute loss for weight  5e-06  0 result 0.978794
 training batch 29 mu var0-0.0512786
compute loss for weight  -5e-06  0 result 0.978794
   --dy = -4.25585e-11 dy_ref = 5.55112e-17
 training batch 30 mu var0-0.0512786
compute loss for weight  1e-05  0 result 0.978794
 training batch 31 mu var0-0.0512786
compute loss for weight  -1e-05  0 result 0.978794
 training batch 32 mu var0-0.0512786
compute loss for weight  5e-06  0 result 0.978794
 training batch 33 mu var0-0.0512786
compute loss for weight  -5e-06  0 result 0.978794
   --dy = -6.29126e-11 dy_ref = -1.38778e-17
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      1.757     -0.1991 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      1.176      0.5435 

 training batch 34 mu var0-0.0512786
compute loss for weight  1.17587  1.17586 result 0.978811
 training batch 35 mu var0-0.0512786
compute loss for weight  1.17585  1.17586 result 0.978776
 training batch 36 mu var0-0.0512786
compute loss for weight  1.17586  1.17586 result 0.978802
 training batch 37 mu var0-0.0512786
compute loss for weight  1.17585  1.17586 result 0.978785
   --dy = 1.75683 dy_ref = 1.75683
 training batch 38 mu var0-0.0512786
compute loss for weight  0.543479  0.543469 result 0.978792
 training batch 39 mu var0-0.0512786
compute loss for weight  0.543459  0.543469 result 0.978796
 training batch 40 mu var0-0.0512786
compute loss for weight  0.543474  0.543469 result 0.978793
 training batch 41 mu var0-0.0512786
compute loss for weight  0.543464  0.543469 result 0.978795
   --dy = -0.199089 dy_ref = -0.199089
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m5.61509e-10[NON-XML-CHAR-0x1B][39m