Execution Time0.97s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: PR-4624-x86_64-fedora30-gcc9-opt (root-fedora30-1.cern.ch) on 2019-11-14 18:11:26

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 var0-0.219287
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     -0.517      0.4915 
   1 |     0.8389     -0.1584 
   2 |     -1.816      0.5994 
   3 |    -0.1236      0.6049 
   4 |    -0.6384      0.6147 
   5 |    -0.9724       1.107 
   6 |      1.296      0.2307 
   7 |    -0.1981      -1.192 
   8 |    -0.1921      0.3711 
   9 |     0.1297     0.04125 

output BN 
output DL feature 0 mean -0.219287	output DL std 0.877573
output DL feature 1 mean 0.27099	output DL std 0.621388
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.3576       0.374 
   1 |      1.271     -0.7282 
   2 |     -1.918       0.557 
   3 |      0.115      0.5664 
   4 |    -0.5034      0.5829 
   5 |    -0.9045       1.418 
   6 |      1.821    -0.06839 
   7 |    0.02539      -2.482 
   8 |    0.03266      0.1699 
   9 |     0.4191     -0.3897 

output BN feature 0 mean -6.10623e-17	output BN std 1.05402
output BN feature 1 mean 1.66533e-17	output BN std 1.05394
Testing weight gradients   for    layer 0
weight gradient for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |      0.302     -0.6996      0.1011      0.1132 
   1 |    -0.1552   -0.003318     -0.6976      0.6991 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |     0.1765     0.06181     -0.8275      0.6507 
   1 |     0.0381     -0.5949      0.2488   -0.008696 

 training batch 2 mu var0-0.219284
compute loss for weight  0.176549  0.176539 result 0.579259
 training batch 3 mu var0-0.219287
compute loss for weight  0.176529  0.176539 result 0.579253
 training batch 4 mu var0-0.219286
compute loss for weight  0.176544  0.176539 result 0.579257
 training batch 5 mu var0-0.219287
compute loss for weight  0.176534  0.176539 result 0.579254
   --dy = 0.301999 dy_ref = 0.301999
 training batch 6 mu var0-0.219288
compute loss for weight  0.0618238  0.0618138 result 0.579249
 training batch 7 mu var0-0.219287
compute loss for weight  0.0618038  0.0618138 result 0.579263
 training batch 8 mu var0-0.219287
compute loss for weight  0.0618188  0.0618138 result 0.579252
 training batch 9 mu var0-0.219287
compute loss for weight  0.0618088  0.0618138 result 0.579259
   --dy = -0.699615 dy_ref = -0.699615
 training batch 10 mu var0-0.219287
compute loss for weight  -0.827508  -0.827518 result 0.579257
 training batch 11 mu var0-0.219287
compute loss for weight  -0.827528  -0.827518 result 0.579255
 training batch 12 mu var0-0.219287
compute loss for weight  -0.827513  -0.827518 result 0.579256
 training batch 13 mu var0-0.219287
compute loss for weight  -0.827523  -0.827518 result 0.579255
   --dy = 0.10113 dy_ref = 0.10113
 training batch 14 mu var0-0.219287
compute loss for weight  0.650754  0.650744 result 0.579257
 training batch 15 mu var0-0.219287
compute loss for weight  0.650734  0.650744 result 0.579255
 training batch 16 mu var0-0.219287
compute loss for weight  0.650749  0.650744 result 0.579256
 training batch 17 mu var0-0.219287
compute loss for weight  0.650739  0.650744 result 0.579255
   --dy = 0.113158 dy_ref = 0.113158
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.1303       1.028 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var0-0.219287
compute loss for weight  1.00001  1 result 0.579257
 training batch 19 mu var0-0.219287
compute loss for weight  0.99999  1 result 0.579255
 training batch 20 mu var0-0.219287
compute loss for weight  1.00001  1 result 0.579256
 training batch 21 mu var0-0.219287
compute loss for weight  0.999995  1 result 0.579255
   --dy = 0.130267 dy_ref = 0.130267
 training batch 22 mu var0-0.219287
compute loss for weight  1.00001  1 result 0.579266
 training batch 23 mu var0-0.219287
compute loss for weight  0.99999  1 result 0.579246
 training batch 24 mu var0-0.219287
compute loss for weight  1.00001  1 result 0.579261
 training batch 25 mu var0-0.219287
compute loss for weight  0.999995  1 result 0.579251
   --dy = 1.02824 dy_ref = 1.02824
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |  1.041e-17   6.939e-18 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var0-0.219287
compute loss for weight  1e-05  0 result 0.579256
 training batch 27 mu var0-0.219287
compute loss for weight  -1e-05  0 result 0.579256
 training batch 28 mu var0-0.219287
compute loss for weight  5e-06  0 result 0.579256
 training batch 29 mu var0-0.219287
compute loss for weight  -5e-06  0 result 0.579256
   --dy = 0 dy_ref = 1.04083e-17
 training batch 30 mu var0-0.219287
compute loss for weight  1e-05  0 result 0.579256
 training batch 31 mu var0-0.219287
compute loss for weight  -1e-05  0 result 0.579256
 training batch 32 mu var0-0.219287
compute loss for weight  5e-06  0 result 0.579256
 training batch 33 mu var0-0.219287
compute loss for weight  -5e-06  0 result 0.579256
   --dy = 1.66533e-11 dy_ref = 6.93889e-18
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.2801        1.26 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.4651      0.8158 

 training batch 34 mu var0-0.219287
compute loss for weight  0.465111  0.465101 result 0.579259
 training batch 35 mu var0-0.219287
compute loss for weight  0.465091  0.465101 result 0.579253
 training batch 36 mu var0-0.219287
compute loss for weight  0.465106  0.465101 result 0.579257
 training batch 37 mu var0-0.219287
compute loss for weight  0.465096  0.465101 result 0.579254
   --dy = 0.280083 dy_ref = 0.280083
 training batch 38 mu var0-0.219287
compute loss for weight  0.815772  0.815762 result 0.579268
 training batch 39 mu var0-0.219287
compute loss for weight  0.815752  0.815762 result 0.579243
 training batch 40 mu var0-0.219287
compute loss for weight  0.815767  0.815762 result 0.579262
 training batch 41 mu var0-0.219287
compute loss for weight  0.815757  0.815762 result 0.57925
   --dy = 1.26047 dy_ref = 1.26047
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m1.68228e-10[NON-XML-CHAR-0x1B][39m