Execution Time0.55s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: master-x86_64-mac1014-clang100-dbg (macphsft17.dyndns.cern.ch) on 2019-11-13 14:24:36

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 var0-0.467432
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     -1.161       1.688 
   1 |     0.8677      0.2698 
   2 |     -1.609      0.9073 
   3 |      1.762      0.1996 
   4 |    -0.7462       1.526 
   5 |      -1.74       3.413 
   6 |      0.432      0.5637 
   7 |    -0.5358      -1.887 
   8 |     -2.412       2.449 
   9 |     0.4683      -0.158 

output BN 
output DL feature 0 mean -0.467432	output DL std 1.3209
output DL feature 1 mean 0.897136	output DL std 1.47616
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.5536      0.5647 
   1 |      1.065      -0.448 
   2 |    -0.9112    0.007241 
   3 |      1.779     -0.4981 
   4 |    -0.2224      0.4489 
   5 |     -1.016       1.796 
   6 |     0.7178     -0.2381 
   7 |   -0.05457      -1.988 
   8 |     -1.551       1.108 
   9 |     0.7467     -0.7534 

output BN feature 0 mean 1.11022e-17	output BN std 1.05406
output BN feature 1 mean 1.22125e-16	output BN std 1.05407
Testing weight gradients   for    layer 0
weight gradient for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |     0.0866     -0.1454     0.01289     0.01188 
   1 |     -1.338        1.04      -1.246     -0.3277 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |    -0.3235     -0.3035      -1.241    0.005695 
   1 |       0.91     -0.7903      0.5652       0.214 

 training batch 2 mu var0-0.467429
compute loss for weight  -0.323455  -0.323465 result 1.50019
 training batch 3 mu var0-0.467432
compute loss for weight  -0.323475  -0.323465 result 1.50019
 training batch 4 mu var0-0.467431
compute loss for weight  -0.32346  -0.323465 result 1.50019
 training batch 5 mu var0-0.467432
compute loss for weight  -0.32347  -0.323465 result 1.50019
   --dy = 0.0866035 dy_ref = 0.0866035
 training batch 6 mu var0-0.467432
compute loss for weight  -0.303469  -0.303479 result 1.50019
 training batch 7 mu var0-0.467432
compute loss for weight  -0.303489  -0.303479 result 1.50019
 training batch 8 mu var0-0.467432
compute loss for weight  -0.303474  -0.303479 result 1.50019
 training batch 9 mu var0-0.467432
compute loss for weight  -0.303484  -0.303479 result 1.50019
   --dy = -0.145447 dy_ref = -0.145447
 training batch 10 mu var0-0.467431
compute loss for weight  -1.2405  -1.24051 result 1.50019
 training batch 11 mu var0-0.467432
compute loss for weight  -1.24052  -1.24051 result 1.50019
 training batch 12 mu var0-0.467432
compute loss for weight  -1.2405  -1.24051 result 1.50019
 training batch 13 mu var0-0.467432
compute loss for weight  -1.24051  -1.24051 result 1.50019
   --dy = 0.0128934 dy_ref = 0.0128934
 training batch 14 mu var0-0.467432
compute loss for weight  0.00570496  0.00569496 result 1.50019
 training batch 15 mu var0-0.467432
compute loss for weight  0.00568496  0.00569496 result 1.50019
 training batch 16 mu var0-0.467432
compute loss for weight  0.00569996  0.00569496 result 1.50019
 training batch 17 mu var0-0.467432
compute loss for weight  0.00568996  0.00569496 result 1.50019
   --dy = 0.0118849 dy_ref = 0.0118849
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      3.144     -0.1439 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var0-0.467432
compute loss for weight  1.00001  1 result 1.50022
 training batch 19 mu var0-0.467432
compute loss for weight  0.99999  1 result 1.50016
 training batch 20 mu var0-0.467432
compute loss for weight  1.00001  1 result 1.50021
 training batch 21 mu var0-0.467432
compute loss for weight  0.999995  1 result 1.50018
   --dy = 3.1443 dy_ref = 3.1443
 training batch 22 mu var0-0.467432
compute loss for weight  1.00001  1 result 1.50019
 training batch 23 mu var0-0.467432
compute loss for weight  0.99999  1 result 1.50019
 training batch 24 mu var0-0.467432
compute loss for weight  1.00001  1 result 1.50019
 training batch 25 mu var0-0.467432
compute loss for weight  0.999995  1 result 1.50019
   --dy = -0.143917 dy_ref = -0.143917
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

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

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var0-0.467432
compute loss for weight  1e-05  0 result 1.50019
 training batch 27 mu var0-0.467432
compute loss for weight  -1e-05  0 result 1.50019
 training batch 28 mu var0-0.467432
compute loss for weight  5e-06  0 result 1.50019
 training batch 29 mu var0-0.467432
compute loss for weight  -5e-06  0 result 1.50019
   --dy = -2.96059e-11 dy_ref = 0
 training batch 30 mu var0-0.467432
compute loss for weight  1e-05  0 result 1.50019
 training batch 31 mu var0-0.467432
compute loss for weight  -1e-05  0 result 1.50019
 training batch 32 mu var0-0.467432
compute loss for weight  5e-06  0 result 1.50019
 training batch 33 mu var0-0.467432
compute loss for weight  -5e-06  0 result 1.50019
   --dy = -2.59052e-11 dy_ref = 0
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     -2.443       1.312 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     -1.287     -0.1097 

 training batch 34 mu var0-0.467432
compute loss for weight  -1.28694  -1.28695 result 1.50017
 training batch 35 mu var0-0.467432
compute loss for weight  -1.28696  -1.28695 result 1.50022
 training batch 36 mu var0-0.467432
compute loss for weight  -1.28695  -1.28695 result 1.50018
 training batch 37 mu var0-0.467432
compute loss for weight  -1.28696  -1.28695 result 1.50021
   --dy = -2.44322 dy_ref = -2.44322
 training batch 38 mu var0-0.467432
compute loss for weight  -0.109653  -0.109663 result 1.50021
 training batch 39 mu var0-0.467432
compute loss for weight  -0.109673  -0.109663 result 1.50018
 training batch 40 mu var0-0.467432
compute loss for weight  -0.109658  -0.109663 result 1.5002
 training batch 41 mu var0-0.467432
compute loss for weight  -0.109668  -0.109663 result 1.50019
   --dy = 1.31235 dy_ref = 1.31235
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][33m3.37951e-09[NON-XML-CHAR-0x1B][39m