Execution Time0.57s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: PR-4623-x86_64-fedora29-gcc8-opt (root-fedora29-2.cern.ch) on 2019-11-14 10:14:09

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 var00.0309903
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.1635     0.01895 
   1 |     0.8324     -0.2559 
   2 |    -0.6624    -0.06221 
   3 |      1.452      -1.566 
   4 |    0.08776     -0.3329 
   5 |     0.1529     -0.4001 
   6 |  -0.003612    -0.01081 
   7 |     -0.709       1.216 
   8 |    -0.9392       0.895 
   9 |      0.263     -0.2552 

output BN 
output DL feature 0 mean 0.0309903	output DL std 0.72579
output DL feature 1 mean -0.0753077	output DL std 0.751438
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.2824      0.1322 
   1 |      1.164     -0.2533 
   2 |     -1.007     0.01837 
   3 |      2.063      -2.091 
   4 |    0.08245     -0.3613 
   5 |      0.177     -0.4555 
   6 |   -0.05025     0.09047 
   7 |     -1.075       1.811 
   8 |     -1.409       1.361 
   9 |      0.337     -0.2523 

output BN feature 0 mean 3.33067e-17	output BN std 1.05398
output BN feature 1 mean -9.4369e-17	output BN std 1.05399
Testing weight gradients   for    layer 0
weight gradient for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |   -0.02979       0.266     -0.1742      0.1131 
   1 |    -0.4158      -5.622      -3.309      -3.271 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |     0.2162     -0.3592      -0.691     -0.1518 
   1 |    0.08004      0.6622      0.2983      0.2089 

 training batch 2 mu var00.0309931
compute loss for weight  0.216219  0.216209 result 2.57876
 training batch 3 mu var00.0309903
compute loss for weight  0.216199  0.216209 result 2.57876
 training batch 4 mu var00.030991
compute loss for weight  0.216214  0.216209 result 2.57876
 training batch 5 mu var00.0309903
compute loss for weight  0.216204  0.216209 result 2.57876
   --dy = -0.0297866 dy_ref = -0.0297866
 training batch 6 mu var00.0309898
compute loss for weight  -0.359208  -0.359218 result 2.57876
 training batch 7 mu var00.0309903
compute loss for weight  -0.359228  -0.359218 result 2.57876
 training batch 8 mu var00.0309901
compute loss for weight  -0.359213  -0.359218 result 2.57876
 training batch 9 mu var00.0309903
compute loss for weight  -0.359223  -0.359218 result 2.57876
   --dy = 0.266004 dy_ref = 0.266004
 training batch 10 mu var00.0309906
compute loss for weight  -0.690952  -0.690962 result 2.57876
 training batch 11 mu var00.0309903
compute loss for weight  -0.690972  -0.690962 result 2.57876
 training batch 12 mu var00.0309905
compute loss for weight  -0.690957  -0.690962 result 2.57876
 training batch 13 mu var00.0309903
compute loss for weight  -0.690967  -0.690962 result 2.57876
   --dy = -0.174187 dy_ref = -0.174187
 training batch 14 mu var00.0309903
compute loss for weight  -0.151813  -0.151823 result 2.57876
 training batch 15 mu var00.0309903
compute loss for weight  -0.151833  -0.151823 result 2.57876
 training batch 16 mu var00.0309903
compute loss for weight  -0.151818  -0.151823 result 2.57876
 training batch 17 mu var00.0309903
compute loss for weight  -0.151828  -0.151823 result 2.57876
   --dy = 0.113146 dy_ref = 0.113146
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      5.619     -0.4618 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var00.0309903
compute loss for weight  1.00001  1 result 2.57882
 training batch 19 mu var00.0309903
compute loss for weight  0.99999  1 result 2.5787
 training batch 20 mu var00.0309903
compute loss for weight  1.00001  1 result 2.57879
 training batch 21 mu var00.0309903
compute loss for weight  0.999995  1 result 2.57873
   --dy = 5.61933 dy_ref = 5.61933
 training batch 22 mu var00.0309903
compute loss for weight  1.00001  1 result 2.57876
 training batch 23 mu var00.0309903
compute loss for weight  0.99999  1 result 2.57877
 training batch 24 mu var00.0309903
compute loss for weight  1.00001  1 result 2.57876
 training batch 25 mu var00.0309903
compute loss for weight  0.999995  1 result 2.57876
   --dy = -0.461812 dy_ref = -0.461812
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |  1.388e-16    4.51e-17 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var00.0309903
compute loss for weight  1e-05  0 result 2.57876
 training batch 27 mu var00.0309903
compute loss for weight  -1e-05  0 result 2.57876
 training batch 28 mu var00.0309903
compute loss for weight  5e-06  0 result 2.57876
 training batch 29 mu var00.0309903
compute loss for weight  -5e-06  0 result 2.57876
   --dy = -9.62193e-11 dy_ref = 1.38778e-16
 training batch 30 mu var00.0309903
compute loss for weight  1e-05  0 result 2.57876
 training batch 31 mu var00.0309903
compute loss for weight  -1e-05  0 result 2.57876
 training batch 32 mu var00.0309903
compute loss for weight  5e-06  0 result 2.57876
 training batch 33 mu var00.0309903
compute loss for weight  -5e-06  0 result 2.57876
   --dy = -7.40149e-12 dy_ref = 4.51028e-17
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     -3.207       2.719 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     -1.752     -0.1698 

 training batch 34 mu var00.0309903
compute loss for weight  -1.75215  -1.75216 result 2.57873
 training batch 35 mu var00.0309903
compute loss for weight  -1.75217  -1.75216 result 2.57879
 training batch 36 mu var00.0309903
compute loss for weight  -1.75215  -1.75216 result 2.57874
 training batch 37 mu var00.0309903
compute loss for weight  -1.75216  -1.75216 result 2.57878
   --dy = -3.20709 dy_ref = -3.20709
 training batch 38 mu var00.0309903
compute loss for weight  -0.169829  -0.169839 result 2.57879
 training batch 39 mu var00.0309903
compute loss for weight  -0.169849  -0.169839 result 2.57873
 training batch 40 mu var00.0309903
compute loss for weight  -0.169834  -0.169839 result 2.57877
 training batch 41 mu var00.0309903
compute loss for weight  -0.169844  -0.169839 result 2.57875
   --dy = 2.71911 dy_ref = 2.71911
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][33m3.79797e-09[NON-XML-CHAR-0x1B][39m