Execution Time0.08s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: master-x86_64-fedora28-gcc8 (sft-fedora-28-1.cern.ch) on 2019-11-14 01:14:20

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 var00.154214
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.4059        1.56 
   1 |    -0.3333      0.5611 
   2 |     0.2893      0.7568 
   3 |    -0.7989      0.4094 
   4 |      0.212       1.425 
   5 |     0.5481       3.255 
   6 |     0.5124     0.04941 
   7 |    -0.2466      -1.375 
   8 |      1.111       2.066 
   9 |    -0.1575     -0.1314 

output BN 
output DL feature 0 mean 0.154214	output DL std 0.546444
output DL feature 1 mean 0.857551	output DL std 1.28843
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.4853      0.5746 
   1 |    -0.9402     -0.2425 
   2 |     0.2605    -0.08246 
   3 |     -1.838     -0.3666 
   4 |     0.1115      0.4639 
   5 |     0.7596       1.962 
   6 |     0.6909     -0.6611 
   7 |    -0.7731      -1.827 
   8 |      1.845      0.9886 
   9 |    -0.6012     -0.8091 

output BN feature 0 mean 1.11022e-17	output BN std 1.0539
output BN feature 1 mean 9.99201e-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 |   -0.01761     0.02205   -0.003966      0.0129 
   1 |  -0.004862   -0.001926    -0.01055    -0.01009 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |    0.07102    -0.05027      0.4124      0.3102 
   1 |      1.033     -0.5674      0.3593    -0.01591 

 training batch 2 mu var00.154217
compute loss for weight  0.0710298  0.0710198 result 0.00998674
 training batch 3 mu var00.154214
compute loss for weight  0.0710098  0.0710198 result 0.00998709
 training batch 4 mu var00.154215
compute loss for weight  0.0710248  0.0710198 result 0.00998682
 training batch 5 mu var00.154214
compute loss for weight  0.0710148  0.0710198 result 0.009987
   --dy = -0.0176134 dy_ref = -0.0176134
 training batch 6 mu var00.154213
compute loss for weight  -0.0502642  -0.0502742 result 0.00998713
 training batch 7 mu var00.154214
compute loss for weight  -0.0502842  -0.0502742 result 0.00998669
 training batch 8 mu var00.154214
compute loss for weight  -0.0502692  -0.0502742 result 0.00998702
 training batch 9 mu var00.154214
compute loss for weight  -0.0502792  -0.0502742 result 0.0099868
   --dy = 0.0220491 dy_ref = 0.0220491
 training batch 10 mu var00.154214
compute loss for weight  0.412361  0.412351 result 0.00998687
 training batch 11 mu var00.154214
compute loss for weight  0.412341  0.412351 result 0.00998695
 training batch 12 mu var00.154214
compute loss for weight  0.412356  0.412351 result 0.00998689
 training batch 13 mu var00.154214
compute loss for weight  0.412346  0.412351 result 0.00998693
   --dy = -0.00396606 dy_ref = -0.00396606
 training batch 14 mu var00.154214
compute loss for weight  0.31022  0.31021 result 0.00998704
 training batch 15 mu var00.154214
compute loss for weight  0.3102  0.31021 result 0.00998678
 training batch 16 mu var00.154214
compute loss for weight  0.310215  0.31021 result 0.00998698
 training batch 17 mu var00.154214
compute loss for weight  0.310205  0.31021 result 0.00998685
   --dy = 0.0128992 dy_ref = 0.0128992
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    0.01783    0.002148 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var00.154214
compute loss for weight  1.00001  1 result 0.00998709
 training batch 19 mu var00.154214
compute loss for weight  0.99999  1 result 0.00998673
 training batch 20 mu var00.154214
compute loss for weight  1.00001  1 result 0.009987
 training batch 21 mu var00.154214
compute loss for weight  0.999995  1 result 0.00998682
   --dy = 0.0178261 dy_ref = 0.0178261
 training batch 22 mu var00.154214
compute loss for weight  1.00001  1 result 0.00998693
 training batch 23 mu var00.154214
compute loss for weight  0.99999  1 result 0.00998689
 training batch 24 mu var00.154214
compute loss for weight  1.00001  1 result 0.00998692
 training batch 25 mu var00.154214
compute loss for weight  0.999995  1 result 0.0099869
   --dy = 0.00214768 dy_ref = 0.00214768
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 | -1.911e-18    7.86e-19 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var00.154214
compute loss for weight  1e-05  0 result 0.00998691
 training batch 27 mu var00.154214
compute loss for weight  -1e-05  0 result 0.00998691
 training batch 28 mu var00.154214
compute loss for weight  5e-06  0 result 0.00998691
 training batch 29 mu var00.154214
compute loss for weight  -5e-06  0 result 0.00998691
   --dy = -2.31296e-13 dy_ref = -1.91091e-18
 training batch 30 mu var00.154214
compute loss for weight  1e-05  0 result 0.00998691
 training batch 31 mu var00.154214
compute loss for weight  -1e-05  0 result 0.00998691
 training batch 32 mu var00.154214
compute loss for weight  5e-06  0 result 0.00998691
 training batch 33 mu var00.154214
compute loss for weight  -5e-06  0 result 0.00998691
   --dy = 2.89121e-14 dy_ref = 7.86047e-19
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.1442     0.02491 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.1236     0.08621 

 training batch 34 mu var00.154214
compute loss for weight  -0.123589  -0.123599 result 0.00998547
 training batch 35 mu var00.154214
compute loss for weight  -0.123609  -0.123599 result 0.00998836
 training batch 36 mu var00.154214
compute loss for weight  -0.123594  -0.123599 result 0.00998619
 training batch 37 mu var00.154214
compute loss for weight  -0.123604  -0.123599 result 0.00998763
   --dy = -0.144226 dy_ref = -0.144226
 training batch 38 mu var00.154214
compute loss for weight  0.0862206  0.0862106 result 0.00998716
 training batch 39 mu var00.154214
compute loss for weight  0.0862006  0.0862106 result 0.00998666
 training batch 40 mu var00.154214
compute loss for weight  0.0862156  0.0862106 result 0.00998704
 training batch 41 mu var00.154214
compute loss for weight  0.0862056  0.0862106 result 0.00998679
   --dy = 0.024912 dy_ref = 0.024912
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m7.31774e-11[NON-XML-CHAR-0x1B][39m