Execution Time0.07s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: master-x86_64-fedora28-gcc8 (sft-fedora-28-1.cern.ch) on 2019-11-16 01:09:06

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 var0-0.00468982
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.1239      0.2481 
   1 |    -0.6416     -0.2909 
   2 |      1.278     -0.3313 
   3 |    -0.6169       -1.78 
   4 |    0.07462     -0.1871 
   5 |    0.03095    -0.02189 
   6 |     -1.655       1.104 
   7 |      1.558      0.2296 
   8 |    0.01321       1.616 
   9 |    -0.2119       -0.26 

output BN 
output DL feature 0 mean -0.00468982	output DL std 0.922482
output DL feature 1 mean 0.0326522	output DL std 0.906706
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      0.147      0.2505 
   1 |    -0.7278     -0.3762 
   2 |      1.465     -0.4231 
   3 |    -0.6995      -2.107 
   4 |    0.09062     -0.2554 
   5 |    0.04072     -0.0634 
   6 |     -1.886       1.245 
   7 |      1.786       0.229 
   8 |    0.02045        1.84 
   9 |    -0.2367     -0.3402 

output BN feature 0 mean -2.22045e-17	output BN std 1.05402
output BN feature 1 mean -2.22045e-17	output BN std 1.05402
Testing weight gradients   for    layer 0
weight gradient for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |    -0.1579     -0.2557      -0.331     -0.5149 
   1 |   -0.06197     -0.4581     -0.4517     0.02544 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |    -0.1143      0.6221      0.6245     -0.6755 
   1 |     0.1162        0.28      0.3521      0.7178 

 training batch 2 mu var0-0.00468701
compute loss for weight  -0.114262  -0.114272 result 0.904757
 training batch 3 mu var0-0.00468982
compute loss for weight  -0.114282  -0.114272 result 0.90476
 training batch 4 mu var0-0.00468912
compute loss for weight  -0.114267  -0.114272 result 0.904757
 training batch 5 mu var0-0.00468982
compute loss for weight  -0.114277  -0.114272 result 0.904759
   --dy = -0.157896 dy_ref = -0.157896
 training batch 6 mu var0-0.00469031
compute loss for weight  0.622066  0.622056 result 0.904756
 training batch 7 mu var0-0.00468982
compute loss for weight  0.622046  0.622056 result 0.904761
 training batch 8 mu var0-0.00469001
compute loss for weight  0.622061  0.622056 result 0.904757
 training batch 9 mu var0-0.00468982
compute loss for weight  0.622051  0.622056 result 0.90476
   --dy = -0.255726 dy_ref = -0.255726
 training batch 10 mu var0-0.00468952
compute loss for weight  0.624527  0.624517 result 0.904755
 training batch 11 mu var0-0.00468982
compute loss for weight  0.624507  0.624517 result 0.904762
 training batch 12 mu var0-0.0046897
compute loss for weight  0.624522  0.624517 result 0.904757
 training batch 13 mu var0-0.00468982
compute loss for weight  0.624512  0.624517 result 0.90476
   --dy = -0.331041 dy_ref = -0.331041
 training batch 14 mu var0-0.00468988
compute loss for weight  -0.675464  -0.675474 result 0.904753
 training batch 15 mu var0-0.00468982
compute loss for weight  -0.675484  -0.675474 result 0.904763
 training batch 16 mu var0-0.00468985
compute loss for weight  -0.675469  -0.675474 result 0.904756
 training batch 17 mu var0-0.00468982
compute loss for weight  -0.675479  -0.675474 result 0.904761
   --dy = -0.51491 dy_ref = -0.51491
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.2718       1.538 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var0-0.00468982
compute loss for weight  1.00001  1 result 0.904761
 training batch 19 mu var0-0.00468982
compute loss for weight  0.99999  1 result 0.904756
 training batch 20 mu var0-0.00468982
compute loss for weight  1.00001  1 result 0.90476
 training batch 21 mu var0-0.00468982
compute loss for weight  0.999995  1 result 0.904757
   --dy = 0.271814 dy_ref = 0.271814
 training batch 22 mu var0-0.00468982
compute loss for weight  1.00001  1 result 0.904774
 training batch 23 mu var0-0.00468982
compute loss for weight  0.99999  1 result 0.904743
 training batch 24 mu var0-0.00468982
compute loss for weight  1.00001  1 result 0.904766
 training batch 25 mu var0-0.00468982
compute loss for weight  0.999995  1 result 0.904751
   --dy = 1.5377 dy_ref = 1.5377
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 | -2.255e-17   2.082e-17 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var0-0.00468982
compute loss for weight  1e-05  0 result 0.904758
 training batch 27 mu var0-0.00468982
compute loss for weight  -1e-05  0 result 0.904758
 training batch 28 mu var0-0.00468982
compute loss for weight  5e-06  0 result 0.904758
 training batch 29 mu var0-0.00468982
compute loss for weight  -5e-06  0 result 0.904758
   --dy = -1.85037e-12 dy_ref = -2.25514e-17
 training batch 30 mu var0-0.00468982
compute loss for weight  1e-05  0 result 0.904758
 training batch 31 mu var0-0.00468982
compute loss for weight  -1e-05  0 result 0.904758
 training batch 32 mu var0-0.00468982
compute loss for weight  5e-06  0 result 0.904758
 training batch 33 mu var0-0.00468982
compute loss for weight  -5e-06  0 result 0.904758
   --dy = 4.44089e-11 dy_ref = 2.08167e-17
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.7987      -1.777 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.3403     -0.8654 

 training batch 34 mu var0-0.00468982
compute loss for weight  0.340324  0.340314 result 0.904766
 training batch 35 mu var0-0.00468982
compute loss for weight  0.340304  0.340314 result 0.90475
 training batch 36 mu var0-0.00468982
compute loss for weight  0.340319  0.340314 result 0.904762
 training batch 37 mu var0-0.00468982
compute loss for weight  0.340309  0.340314 result 0.904754
   --dy = 0.798715 dy_ref = 0.798715
 training batch 38 mu var0-0.00468982
compute loss for weight  -0.865348  -0.865358 result 0.90474
 training batch 39 mu var0-0.00468982
compute loss for weight  -0.865368  -0.865358 result 0.904776
 training batch 40 mu var0-0.00468982
compute loss for weight  -0.865353  -0.865358 result 0.904749
 training batch 41 mu var0-0.00468982
compute loss for weight  -0.865363  -0.865358 result 0.904767
   --dy = -1.77696 dy_ref = -1.77696
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m4.44089e-11[NON-XML-CHAR-0x1B][39m