Execution Time0.60s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: PR-4279-x86_64-fedora30-gcc9-opt (root-fedora30-1.cern.ch) on 2019-11-14 21:01:24

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 var00.29143
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.4115       0.462 
   1 |     0.8386      0.2891 
   2 |    -0.6336       1.044 
   3 |     0.6079       2.185 
   4 |     0.3892      0.9301 
   5 |      1.062       1.532 
   6 |      0.611      -1.227 
   7 |     -0.977      -1.042 
   8 |     0.5183     -0.8825 
   9 |    0.08605      0.2237 

output BN 
output DL feature 0 mean 0.29143	output DL std 0.639297
output DL feature 1 mean 0.351451	output DL std 1.13331
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.1979      0.1028 
   1 |     0.9021    -0.05798 
   2 |     -1.525      0.6443 
   3 |     0.5217       1.705 
   4 |     0.1612      0.5382 
   5 |      1.271       1.098 
   6 |     0.5269      -1.468 
   7 |     -2.091      -1.296 
   8 |     0.3741      -1.148 
   9 |    -0.3386     -0.1188 

output BN feature 0 mean 3.88578e-17	output BN std 1.05395
output BN feature 1 mean -2.77556e-17	output BN std 1.05405
Testing weight gradients   for    layer 0
weight gradient for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |  -0.002406      0.2479     0.01573      0.4256 
   1 |     -0.151      0.2303     0.04009    -0.02763 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |     0.5732     -0.3693     -0.4025      0.2333 
   1 |      0.178     -0.6778    -0.04572     -0.8405 

 training batch 2 mu var00.291433
compute loss for weight  0.573243  0.573233 result 0.234446
 training batch 3 mu var00.29143
compute loss for weight  0.573223  0.573233 result 0.234446
 training batch 4 mu var00.291431
compute loss for weight  0.573238  0.573233 result 0.234446
 training batch 5 mu var00.29143
compute loss for weight  0.573228  0.573233 result 0.234446
   --dy = -0.00240586 dy_ref = -0.00240586
 training batch 6 mu var00.291429
compute loss for weight  -0.369262  -0.369272 result 0.234449
 training batch 7 mu var00.29143
compute loss for weight  -0.369282  -0.369272 result 0.234444
 training batch 8 mu var00.29143
compute loss for weight  -0.369267  -0.369272 result 0.234448
 training batch 9 mu var00.29143
compute loss for weight  -0.369277  -0.369272 result 0.234445
   --dy = 0.247854 dy_ref = 0.247854
 training batch 10 mu var00.29143
compute loss for weight  -0.402487  -0.402497 result 0.234447
 training batch 11 mu var00.29143
compute loss for weight  -0.402507  -0.402497 result 0.234446
 training batch 12 mu var00.29143
compute loss for weight  -0.402492  -0.402497 result 0.234447
 training batch 13 mu var00.29143
compute loss for weight  -0.402502  -0.402497 result 0.234446
   --dy = 0.0157303 dy_ref = 0.0157303
 training batch 14 mu var00.29143
compute loss for weight  0.233351  0.233341 result 0.234451
 training batch 15 mu var00.29143
compute loss for weight  0.233331  0.233341 result 0.234442
 training batch 16 mu var00.29143
compute loss for weight  0.233346  0.233341 result 0.234449
 training batch 17 mu var00.29143
compute loss for weight  0.233336  0.233341 result 0.234444
   --dy = 0.425619 dy_ref = 0.425619
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.2864      0.1825 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var00.29143
compute loss for weight  1.00001  1 result 0.234449
 training batch 19 mu var00.29143
compute loss for weight  0.99999  1 result 0.234444
 training batch 20 mu var00.29143
compute loss for weight  1.00001  1 result 0.234448
 training batch 21 mu var00.29143
compute loss for weight  0.999995  1 result 0.234445
   --dy = 0.286416 dy_ref = 0.286416
 training batch 22 mu var00.29143
compute loss for weight  1.00001  1 result 0.234448
 training batch 23 mu var00.29143
compute loss for weight  0.99999  1 result 0.234445
 training batch 24 mu var00.29143
compute loss for weight  1.00001  1 result 0.234447
 training batch 25 mu var00.29143
compute loss for weight  0.999995  1 result 0.234446
   --dy = 0.182477 dy_ref = 0.182477
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |  2.776e-17  -2.168e-17 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var00.29143
compute loss for weight  1e-05  0 result 0.234446
 training batch 27 mu var00.29143
compute loss for weight  -1e-05  0 result 0.234446
 training batch 28 mu var00.29143
compute loss for weight  5e-06  0 result 0.234446
 training batch 29 mu var00.29143
compute loss for weight  -5e-06  0 result 0.234446
   --dy = -7.40149e-12 dy_ref = 2.77556e-17
 training batch 30 mu var00.29143
compute loss for weight  1e-05  0 result 0.234446
 training batch 31 mu var00.29143
compute loss for weight  -1e-05  0 result 0.234446
 training batch 32 mu var00.29143
compute loss for weight  5e-06  0 result 0.234446
 training batch 33 mu var00.29143
compute loss for weight  -5e-06  0 result 0.234446
   --dy = 3.70074e-12 dy_ref = -2.1684e-17
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.6563     -0.4905 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.4364      -0.372 

 training batch 34 mu var00.29143
compute loss for weight  0.436389  0.436379 result 0.234453
 training batch 35 mu var00.29143
compute loss for weight  0.436369  0.436379 result 0.23444
 training batch 36 mu var00.29143
compute loss for weight  0.436384  0.436379 result 0.23445
 training batch 37 mu var00.29143
compute loss for weight  0.436374  0.436379 result 0.234443
   --dy = 0.656348 dy_ref = 0.656348
 training batch 38 mu var00.29143
compute loss for weight  -0.372034  -0.372044 result 0.234442
 training batch 39 mu var00.29143
compute loss for weight  -0.372054  -0.372044 result 0.234451
 training batch 40 mu var00.29143
compute loss for weight  -0.372039  -0.372044 result 0.234444
 training batch 41 mu var00.29143
compute loss for weight  -0.372049  -0.372044 result 0.234449
   --dy = -0.490471 dy_ref = -0.490471
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][33m5.08923e-09[NON-XML-CHAR-0x1B][39m