Execution Time0.59s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: PR-4279-x86_64-fedora29-gcc8-opt (root-fedora29-3.cern.ch) on 2019-11-14 21:03:34

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 var00.0282926
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    0.06953     -0.1789 
   1 |    -0.7031     -0.8985 
   2 |     0.9236       1.091 
   3 |     0.8003      0.4444 
   4 |      0.366     0.09004 
   5 |     0.2617     -0.3368 
   6 |    0.06473     -0.5549 
   7 |     -1.142     -0.1922 
   8 |    -0.4701     -0.8004 
   9 |      0.112     0.05246 

output BN 
output DL feature 0 mean 0.0282926	output DL std 0.643602
output DL feature 1 mean -0.128421	output DL std 0.59359
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    0.06753    -0.08971 
   1 |     -1.198      -1.367 
   2 |      1.466       2.164 
   3 |      1.264       1.017 
   4 |      0.553      0.3879 
   5 |     0.3822     -0.3699 
   6 |    0.05966     -0.7572 
   7 |     -1.916     -0.1133 
   8 |    -0.8162      -1.193 
   9 |      0.137      0.3212 

output BN feature 0 mean -6.38378e-17	output BN std 1.05395
output BN feature 1 mean 1.11022e-17	output BN std 1.05393
Testing weight gradients   for    layer 0
weight gradient for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |     0.2045     -0.2165    -0.01089      0.2433 
   1 |    -0.2175        0.37    -0.01089      -0.207 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |    -0.5114     -0.6657      0.3666     -0.1464 
   1 |    -0.6521     -0.2556      0.4478     -0.3735 

 training batch 2 mu var00.0282954
compute loss for weight  -0.511413  -0.511423 result 0.233415
 training batch 3 mu var00.0282926
compute loss for weight  -0.511433  -0.511423 result 0.233411
 training batch 4 mu var00.0282933
compute loss for weight  -0.511418  -0.511423 result 0.233414
 training batch 5 mu var00.0282926
compute loss for weight  -0.511428  -0.511423 result 0.233412
   --dy = 0.20449 dy_ref = 0.20449
 training batch 6 mu var00.0282921
compute loss for weight  -0.665716  -0.665726 result 0.233411
 training batch 7 mu var00.0282926
compute loss for weight  -0.665736  -0.665726 result 0.233415
 training batch 8 mu var00.0282924
compute loss for weight  -0.665721  -0.665726 result 0.233412
 training batch 9 mu var00.0282926
compute loss for weight  -0.665731  -0.665726 result 0.233414
   --dy = -0.216541 dy_ref = -0.216541
 training batch 10 mu var00.0282929
compute loss for weight  0.366573  0.366563 result 0.233413
 training batch 11 mu var00.0282926
compute loss for weight  0.366553  0.366563 result 0.233413
 training batch 12 mu var00.0282928
compute loss for weight  0.366568  0.366563 result 0.233413
 training batch 13 mu var00.0282926
compute loss for weight  0.366558  0.366563 result 0.233413
   --dy = -0.010891 dy_ref = -0.010891
 training batch 14 mu var00.0282926
compute loss for weight  -0.146357  -0.146367 result 0.233415
 training batch 15 mu var00.0282926
compute loss for weight  -0.146377  -0.146367 result 0.23341
 training batch 16 mu var00.0282926
compute loss for weight  -0.146362  -0.146367 result 0.233414
 training batch 17 mu var00.0282926
compute loss for weight  -0.146372  -0.146367 result 0.233412
   --dy = 0.243312 dy_ref = 0.243312
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.1102      0.5771 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var00.0282926
compute loss for weight  1.00001  1 result 0.233412
 training batch 19 mu var00.0282926
compute loss for weight  0.99999  1 result 0.233414
 training batch 20 mu var00.0282926
compute loss for weight  1.00001  1 result 0.233412
 training batch 21 mu var00.0282926
compute loss for weight  0.999995  1 result 0.233413
   --dy = -0.110234 dy_ref = -0.110234
 training batch 22 mu var00.0282926
compute loss for weight  1.00001  1 result 0.233419
 training batch 23 mu var00.0282926
compute loss for weight  0.99999  1 result 0.233407
 training batch 24 mu var00.0282926
compute loss for weight  1.00001  1 result 0.233416
 training batch 25 mu var00.0282926
compute loss for weight  0.999995  1 result 0.23341
   --dy = 0.57706 dy_ref = 0.57706
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 | -3.469e-18  -6.939e-18 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var00.0282926
compute loss for weight  1e-05  0 result 0.233413
 training batch 27 mu var00.0282926
compute loss for weight  -1e-05  0 result 0.233413
 training batch 28 mu var00.0282926
compute loss for weight  5e-06  0 result 0.233413
 training batch 29 mu var00.0282926
compute loss for weight  -5e-06  0 result 0.233413
   --dy = -8.32667e-12 dy_ref = -3.46945e-18
 training batch 30 mu var00.0282926
compute loss for weight  1e-05  0 result 0.233413
 training batch 31 mu var00.0282926
compute loss for weight  -1e-05  0 result 0.233413
 training batch 32 mu var00.0282926
compute loss for weight  5e-06  0 result 0.233413
 training batch 33 mu var00.0282926
compute loss for weight  -5e-06  0 result 0.233413
   --dy = -8.32667e-12 dy_ref = -6.93889e-18
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.3561      0.8704 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.3096       0.663 

 training batch 34 mu var00.0282926
compute loss for weight  -0.30955  -0.30956 result 0.233416
 training batch 35 mu var00.0282926
compute loss for weight  -0.30957  -0.30956 result 0.233409
 training batch 36 mu var00.0282926
compute loss for weight  -0.309555  -0.30956 result 0.233415
 training batch 37 mu var00.0282926
compute loss for weight  -0.309565  -0.30956 result 0.233411
   --dy = 0.3561 dy_ref = 0.3561
 training batch 38 mu var00.0282926
compute loss for weight  0.663024  0.663014 result 0.233421
 training batch 39 mu var00.0282926
compute loss for weight  0.663004  0.663014 result 0.233404
 training batch 40 mu var00.0282926
compute loss for weight  0.663019  0.663014 result 0.233417
 training batch 41 mu var00.0282926
compute loss for weight  0.663009  0.663014 result 0.233408
   --dy = 0.870359 dy_ref = 0.870359
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m7.58983e-11[NON-XML-CHAR-0x1B][39m