Execution Time0.60s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: PR-4624-x86_64-fedora27-gcc7-opt (sft-fedora-27-2.cern.ch) on 2019-11-14 18:10:17
Repository revision: 68fa649de1099da49c4a656bb6224d1fde791c80

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 var0-0.046888
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |  -0.009047      0.1342 
   1 |    0.04982     -0.2472 
   2 |    -0.6897      0.1768 
   3 |     -1.049     -0.1043 
   4 |    -0.2879      0.1275 
   5 |    -0.2985      0.2035 
   6 |      1.011      0.4054 
   7 |    0.05466     -0.4906 
   8 |     0.8607      0.3102 
   9 |    -0.1104    -0.01586 

output BN 
output DL feature 0 mean -0.046888	output DL std 0.624195
output DL feature 1 mean 0.0499612	output DL std 0.269889
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    0.06389      0.3288 
   1 |     0.1633       -1.16 
   2 |     -1.085       0.495 
   3 |     -1.692     -0.6022 
   4 |    -0.4069      0.3027 
   5 |    -0.4249      0.5992 
   6 |      1.786       1.387 
   7 |     0.1715       -2.11 
   8 |      1.532       1.016 
   9 |    -0.1073     -0.2569 

output BN feature 0 mean 5.55112e-17	output BN std 1.05394
output BN feature 1 mean -2.22045e-17	output BN std 1.05329
Testing weight gradients   for    layer 0
weight gradient for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |    -0.5656       1.511      -1.031     -0.3655 
   1 |      -3.91       4.164      -6.495      -10.42 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |    0.09991       0.166    -0.02393      0.5979 
   1 |    -0.1002     -0.2223      0.1826      0.1703 

 training batch 2 mu var0-0.0468852
compute loss for weight  0.0999151  0.0999051 result 0.992018
 training batch 3 mu var0-0.046888
compute loss for weight  0.0998951  0.0999051 result 0.992029
 training batch 4 mu var0-0.0468873
compute loss for weight  0.0999101  0.0999051 result 0.992021
 training batch 5 mu var0-0.046888
compute loss for weight  0.0999001  0.0999051 result 0.992026
   --dy = -0.565617 dy_ref = -0.565617
 training batch 6 mu var0-0.0468885
compute loss for weight  0.166007  0.165997 result 0.992039
 training batch 7 mu var0-0.046888
compute loss for weight  0.165987  0.165997 result 0.992008
 training batch 8 mu var0-0.0468882
compute loss for weight  0.166002  0.165997 result 0.992031
 training batch 9 mu var0-0.046888
compute loss for weight  0.165992  0.165997 result 0.992016
   --dy = 1.51107 dy_ref = 1.51107
 training batch 10 mu var0-0.0468877
compute loss for weight  -0.0239179  -0.0239279 result 0.992013
 training batch 11 mu var0-0.046888
compute loss for weight  -0.0239379  -0.0239279 result 0.992034
 training batch 12 mu var0-0.0468879
compute loss for weight  -0.0239229  -0.0239279 result 0.992018
 training batch 13 mu var0-0.046888
compute loss for weight  -0.0239329  -0.0239279 result 0.992029
   --dy = -1.03068 dy_ref = -1.03068
 training batch 14 mu var0-0.0468881
compute loss for weight  0.597937  0.597927 result 0.99202
 training batch 15 mu var0-0.046888
compute loss for weight  0.597917  0.597927 result 0.992027
 training batch 16 mu var0-0.046888
compute loss for weight  0.597932  0.597927 result 0.992022
 training batch 17 mu var0-0.046888
compute loss for weight  0.597922  0.597927 result 0.992025
   --dy = -0.365522 dy_ref = -0.365522
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      1.515      0.4694 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var0-0.046888
compute loss for weight  1.00001  1 result 0.992039
 training batch 19 mu var0-0.046888
compute loss for weight  0.99999  1 result 0.992008
 training batch 20 mu var0-0.046888
compute loss for weight  1.00001  1 result 0.992031
 training batch 21 mu var0-0.046888
compute loss for weight  0.999995  1 result 0.992016
   --dy = 1.51461 dy_ref = 1.51461
 training batch 22 mu var0-0.046888
compute loss for weight  1.00001  1 result 0.992028
 training batch 23 mu var0-0.046888
compute loss for weight  0.99999  1 result 0.992019
 training batch 24 mu var0-0.046888
compute loss for weight  1.00001  1 result 0.992026
 training batch 25 mu var0-0.046888
compute loss for weight  0.999995  1 result 0.992021
   --dy = 0.469439 dy_ref = 0.469439
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |  1.457e-16  -8.327e-17 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var0-0.046888
compute loss for weight  1e-05  0 result 0.992024
 training batch 27 mu var0-0.046888
compute loss for weight  -1e-05  0 result 0.992024
 training batch 28 mu var0-0.046888
compute loss for weight  5e-06  0 result 0.992024
 training batch 29 mu var0-0.046888
compute loss for weight  -5e-06  0 result 0.992024
   --dy = 0 dy_ref = 1.45717e-16
 training batch 30 mu var0-0.046888
compute loss for weight  1e-05  0 result 0.992024
 training batch 31 mu var0-0.046888
compute loss for weight  -1e-05  0 result 0.992024
 training batch 32 mu var0-0.046888
compute loss for weight  5e-06  0 result 0.992024
 training batch 33 mu var0-0.046888
compute loss for weight  -5e-06  0 result 0.992024
   --dy = 1.66533e-11 dy_ref = -8.32667e-17
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      1.504      -0.669 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      1.007     -0.7017 

 training batch 34 mu var0-0.046888
compute loss for weight  1.00725  1.00724 result 0.992039
 training batch 35 mu var0-0.046888
compute loss for weight  1.00723  1.00724 result 0.992008
 training batch 36 mu var0-0.046888
compute loss for weight  1.00725  1.00724 result 0.992031
 training batch 37 mu var0-0.046888
compute loss for weight  1.00724  1.00724 result 0.992016
   --dy = 1.50372 dy_ref = 1.50372
 training batch 38 mu var0-0.046888
compute loss for weight  -0.70171  -0.70172 result 0.992017
 training batch 39 mu var0-0.046888
compute loss for weight  -0.70173  -0.70172 result 0.99203
 training batch 40 mu var0-0.046888
compute loss for weight  -0.701715  -0.70172 result 0.99202
 training batch 41 mu var0-0.046888
compute loss for weight  -0.701725  -0.70172 result 0.992027
   --dy = -0.668983 dy_ref = -0.668983
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m9.29722e-11[NON-XML-CHAR-0x1B][39m