Execution Time0.53s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: PR-4625-x86_64-fedora29-gcc8-opt (root-fedora29-2.cern.ch) on 2019-11-15 11:03:26

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 var0-0.246224
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.3343      0.6001 
   1 |     -1.143      -2.054 
   2 |      1.323           2 
   3 |   -0.05188      -1.804 
   4 |    -0.1301      0.3966 
   5 |    -0.7945      0.4678 
   6 |     -1.101      0.4692 
   7 |     0.7147      -0.337 
   8 |    -0.9073       1.673 
   9 |   -0.03835     -0.3681 

output BN 
output DL feature 0 mean -0.246224	output DL std 0.800043
output DL feature 1 mean 0.104328	output DL std 1.30693
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     -0.116      0.3999 
   1 |     -1.182      -1.741 
   2 |      2.068       1.529 
   3 |      0.256      -1.539 
   4 |      0.153      0.2357 
   5 |    -0.7223      0.2931 
   6 |     -1.125      0.2943 
   7 |      1.266     -0.3559 
   8 |    -0.8709       1.265 
   9 |     0.2739      -0.381 

output BN feature 0 mean -5.55112e-18	output BN std 1.054
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 |   0.004104  -0.0005148     0.01076    0.006029 
   1 |  -0.004125    0.002138   -0.001614   -0.005961 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |    -0.7628      0.1469      0.6075     -0.5526 
   1 |    -0.7285     -0.1965       1.601      0.3063 

 training batch 2 mu var0-0.246221
compute loss for weight  -0.762831  -0.762841 result 0.231042
 training batch 3 mu var0-0.246224
compute loss for weight  -0.762851  -0.762841 result 0.231042
 training batch 4 mu var0-0.246223
compute loss for weight  -0.762836  -0.762841 result 0.231042
 training batch 5 mu var0-0.246224
compute loss for weight  -0.762846  -0.762841 result 0.231042
   --dy = 0.00410416 dy_ref = 0.00410416
 training batch 6 mu var0-0.246224
compute loss for weight  0.146863  0.146853 result 0.231042
 training batch 7 mu var0-0.246224
compute loss for weight  0.146843  0.146853 result 0.231042
 training batch 8 mu var0-0.246224
compute loss for weight  0.146858  0.146853 result 0.231042
 training batch 9 mu var0-0.246224
compute loss for weight  0.146848  0.146853 result 0.231042
   --dy = -0.000514796 dy_ref = -0.000514796
 training batch 10 mu var0-0.246223
compute loss for weight  0.607557  0.607547 result 0.231042
 training batch 11 mu var0-0.246224
compute loss for weight  0.607537  0.607547 result 0.231042
 training batch 12 mu var0-0.246223
compute loss for weight  0.607552  0.607547 result 0.231042
 training batch 13 mu var0-0.246224
compute loss for weight  0.607542  0.607547 result 0.231042
   --dy = 0.0107623 dy_ref = 0.0107623
 training batch 14 mu var0-0.246224
compute loss for weight  -0.552571  -0.552581 result 0.231042
 training batch 15 mu var0-0.246224
compute loss for weight  -0.552591  -0.552581 result 0.231042
 training batch 16 mu var0-0.246224
compute loss for weight  -0.552576  -0.552581 result 0.231042
 training batch 17 mu var0-0.246224
compute loss for weight  -0.552586  -0.552581 result 0.231042
   --dy = 0.00602927 dy_ref = 0.00602927
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    0.00303      0.4591 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var0-0.246224
compute loss for weight  1.00001  1 result 0.231042
 training batch 19 mu var0-0.246224
compute loss for weight  0.99999  1 result 0.231042
 training batch 20 mu var0-0.246224
compute loss for weight  1.00001  1 result 0.231042
 training batch 21 mu var0-0.246224
compute loss for weight  0.999995  1 result 0.231042
   --dy = 0.00303021 dy_ref = 0.00303021
 training batch 22 mu var0-0.246224
compute loss for weight  1.00001  1 result 0.231046
 training batch 23 mu var0-0.246224
compute loss for weight  0.99999  1 result 0.231037
 training batch 24 mu var0-0.246224
compute loss for weight  1.00001  1 result 0.231044
 training batch 25 mu var0-0.246224
compute loss for weight  0.999995  1 result 0.231039
   --dy = 0.459053 dy_ref = 0.459053
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 | -6.505e-19  -2.776e-17 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var0-0.246224
compute loss for weight  1e-05  0 result 0.231042
 training batch 27 mu var0-0.246224
compute loss for weight  -1e-05  0 result 0.231042
 training batch 28 mu var0-0.246224
compute loss for weight  5e-06  0 result 0.231042
 training batch 29 mu var0-0.246224
compute loss for weight  -5e-06  0 result 0.231042
   --dy = 0 dy_ref = -6.50521e-19
 training batch 30 mu var0-0.246224
compute loss for weight  1e-05  0 result 0.231042
 training batch 31 mu var0-0.246224
compute loss for weight  -1e-05  0 result 0.231042
 training batch 32 mu var0-0.246224
compute loss for weight  5e-06  0 result 0.231042
 training batch 33 mu var0-0.246224
compute loss for weight  -5e-06  0 result 0.231042
   --dy = 6.4763e-12 dy_ref = -2.77556e-17
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.2721      0.9611 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    0.01114      0.4777 

 training batch 34 mu var0-0.246224
compute loss for weight  0.0111463  0.0111363 result 0.231044
 training batch 35 mu var0-0.246224
compute loss for weight  0.0111263  0.0111363 result 0.231039
 training batch 36 mu var0-0.246224
compute loss for weight  0.0111413  0.0111363 result 0.231043
 training batch 37 mu var0-0.246224
compute loss for weight  0.0111313  0.0111363 result 0.23104
   --dy = 0.272102 dy_ref = 0.272102
 training batch 38 mu var0-0.246224
compute loss for weight  0.477661  0.477651 result 0.231051
 training batch 39 mu var0-0.246224
compute loss for weight  0.477641  0.477651 result 0.231032
 training batch 40 mu var0-0.246224
compute loss for weight  0.477656  0.477651 result 0.231047
 training batch 41 mu var0-0.246224
compute loss for weight  0.477646  0.477651 result 0.231037
   --dy = 0.961064 dy_ref = 0.961064
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][33m3.61488e-09[NON-XML-CHAR-0x1B][39m