Execution Time0.57s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: PR-4624-x86_64-fedora29-gcc8-opt (root-fedora29-2.cern.ch) on 2019-11-14 19:02:58

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 var0-0.099238
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.2351       1.581 
   1 |     0.6116      0.6541 
   2 |    -0.3065      0.2682 
   3 |    -0.3093     -0.6095 
   4 |    -0.3635       1.171 
   5 |    -0.4534       3.036 
   6 |     -1.448      0.5201 
   7 |      2.011     -0.8802 
   8 |    -0.4113       2.823 
   9 |   -0.08765     -0.2694 

output BN 
output DL feature 0 mean -0.099238	output DL std 0.892949
output DL feature 1 mean 0.82933	output DL std 1.34014
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.1604      0.5908 
   1 |     0.8391     -0.1378 
   2 |    -0.2446     -0.4413 
   3 |    -0.2479      -1.132 
   4 |    -0.3119      0.2687 
   5 |    -0.4181       1.735 
   6 |     -1.592     -0.2432 
   7 |      2.491      -1.345 
   8 |    -0.3683       1.568 
   9 |    0.01368     -0.8642 

output BN feature 0 mean -6.55725e-17	output BN std 1.05402
output BN feature 1 mean 4.44089e-17	output BN std 1.05406
Testing weight gradients   for    layer 0
weight gradient for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |     0.2454   -0.004012      0.1959     0.06464 
   1 |    0.04987      0.1365     0.02639    -0.05026 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |       0.41      0.9455     -0.3192     -0.5307 
   1 |       1.21     -0.2212      0.3661      0.3467 

 training batch 2 mu var0-0.0992351
compute loss for weight  0.410049  0.410039 result 0.423523
 training batch 3 mu var0-0.099238
compute loss for weight  0.410029  0.410039 result 0.423519
 training batch 4 mu var0-0.0992373
compute loss for weight  0.410044  0.410039 result 0.423522
 training batch 5 mu var0-0.099238
compute loss for weight  0.410034  0.410039 result 0.42352
   --dy = 0.245389 dy_ref = 0.245389
 training batch 6 mu var0-0.0992385
compute loss for weight  0.945489  0.945479 result 0.423521
 training batch 7 mu var0-0.099238
compute loss for weight  0.945469  0.945479 result 0.423521
 training batch 8 mu var0-0.0992381
compute loss for weight  0.945484  0.945479 result 0.423521
 training batch 9 mu var0-0.099238
compute loss for weight  0.945474  0.945479 result 0.423521
   --dy = -0.00401152 dy_ref = -0.00401152
 training batch 10 mu var0-0.0992377
compute loss for weight  -0.319239  -0.319249 result 0.423523
 training batch 11 mu var0-0.099238
compute loss for weight  -0.319259  -0.319249 result 0.423519
 training batch 12 mu var0-0.0992378
compute loss for weight  -0.319244  -0.319249 result 0.423522
 training batch 13 mu var0-0.099238
compute loss for weight  -0.319254  -0.319249 result 0.42352
   --dy = 0.195856 dy_ref = 0.195856
 training batch 14 mu var0-0.099238
compute loss for weight  -0.530709  -0.530719 result 0.423522
 training batch 15 mu var0-0.099238
compute loss for weight  -0.530729  -0.530719 result 0.42352
 training batch 16 mu var0-0.099238
compute loss for weight  -0.530714  -0.530719 result 0.423521
 training batch 17 mu var0-0.099238
compute loss for weight  -0.530724  -0.530719 result 0.423521
   --dy = 0.0646372 dy_ref = 0.0646372
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |   -0.03349      0.8805 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var0-0.099238
compute loss for weight  1.00001  1 result 0.423521
 training batch 19 mu var0-0.099238
compute loss for weight  0.99999  1 result 0.423521
 training batch 20 mu var0-0.099238
compute loss for weight  1.00001  1 result 0.423521
 training batch 21 mu var0-0.099238
compute loss for weight  0.999995  1 result 0.423521
   --dy = -0.0334932 dy_ref = -0.0334932
 training batch 22 mu var0-0.099238
compute loss for weight  1.00001  1 result 0.42353
 training batch 23 mu var0-0.099238
compute loss for weight  0.99999  1 result 0.423512
 training batch 24 mu var0-0.099238
compute loss for weight  1.00001  1 result 0.423525
 training batch 25 mu var0-0.099238
compute loss for weight  0.999995  1 result 0.423517
   --dy = 0.880535 dy_ref = 0.880535
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |  1.388e-17   8.327e-17 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var0-0.099238
compute loss for weight  1e-05  0 result 0.423521
 training batch 27 mu var0-0.099238
compute loss for weight  -1e-05  0 result 0.423521
 training batch 28 mu var0-0.099238
compute loss for weight  5e-06  0 result 0.423521
 training batch 29 mu var0-0.099238
compute loss for weight  -5e-06  0 result 0.423521
   --dy = 9.25186e-13 dy_ref = 1.38778e-17
 training batch 30 mu var0-0.099238
compute loss for weight  1e-05  0 result 0.423521
 training batch 31 mu var0-0.099238
compute loss for weight  -1e-05  0 result 0.423521
 training batch 32 mu var0-0.099238
compute loss for weight  5e-06  0 result 0.423521
 training batch 33 mu var0-0.099238
compute loss for weight  -5e-06  0 result 0.423521
   --dy = 7.40149e-12 dy_ref = 8.32667e-17
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.1506       1.237 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.2224      0.7117 

 training batch 34 mu var0-0.099238
compute loss for weight  0.222376  0.222366 result 0.42352
 training batch 35 mu var0-0.099238
compute loss for weight  0.222356  0.222366 result 0.423523
 training batch 36 mu var0-0.099238
compute loss for weight  0.222371  0.222366 result 0.42352
 training batch 37 mu var0-0.099238
compute loss for weight  0.222361  0.222366 result 0.423522
   --dy = -0.150622 dy_ref = -0.150622
 training batch 38 mu var0-0.099238
compute loss for weight  0.711688  0.711678 result 0.423533
 training batch 39 mu var0-0.099238
compute loss for weight  0.711668  0.711678 result 0.423509
 training batch 40 mu var0-0.099238
compute loss for weight  0.711683  0.711678 result 0.423527
 training batch 41 mu var0-0.099238
compute loss for weight  0.711673  0.711678 result 0.423515
   --dy = 1.23727 dy_ref = 1.23727
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m6.93735e-10[NON-XML-CHAR-0x1B][39m