Execution Time0.24s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: master-x86_64-centos7-gcc62-opt-no-rt-cxxmodules (olhswep09.cern.ch) on 2019-11-14 01:21:50
Repository revision: 32b17abcda23e44b64218a42d0ca69cb30cda7e0

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 var0-0.160443
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.3827     -0.9541 
   1 |    -0.4374      0.4276 
   2 |      1.558     -0.8698 
   3 |      1.516      0.6221 
   4 |     0.1286     -0.7649 
   5 |     -0.377      -1.667 
   6 |     -2.739     -0.5055 
   7 |      1.174      0.8788 
   8 |     -2.183      -1.763 
   9 |     0.1384      0.2037 

output BN 
output DL feature 0 mean -0.160443	output DL std 1.43871
output DL feature 1 mean -0.439149	output DL std 0.933767
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.1629     -0.5812 
   1 |    -0.2029      0.9783 
   2 |      1.259     -0.4861 
   3 |      1.228       1.198 
   4 |     0.2118     -0.3677 
   5 |    -0.1586      -1.386 
   6 |     -1.889    -0.07484 
   7 |     0.9773       1.488 
   8 |     -1.482      -1.494 
   9 |     0.2189      0.7256 

output BN feature 0 mean 2.77556e-18	output BN std 1.05406
output BN feature 1 mean -1.11022e-17	output BN std 1.05403
Testing weight gradients   for    layer 0
weight gradient for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |    -0.1146     0.08192     -0.1803     0.02829 
   1 |     0.4927     -0.3359      0.7428    -0.01841 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |    -0.4604      0.2267      0.1694      -1.441 
   1 |    -0.2548      0.3544     -0.7079     -0.2667 

 training batch 2 mu var0-0.16044
compute loss for weight  -0.460432  -0.460442 result 0.537244
 training batch 3 mu var0-0.160443
compute loss for weight  -0.460452  -0.460442 result 0.537246
 training batch 4 mu var0-0.160443
compute loss for weight  -0.460437  -0.460442 result 0.537245
 training batch 5 mu var0-0.160443
compute loss for weight  -0.460447  -0.460442 result 0.537246
   --dy = -0.114574 dy_ref = -0.114574
 training batch 6 mu var0-0.160444
compute loss for weight  0.226672  0.226662 result 0.537246
 training batch 7 mu var0-0.160443
compute loss for weight  0.226652  0.226662 result 0.537244
 training batch 8 mu var0-0.160443
compute loss for weight  0.226667  0.226662 result 0.537246
 training batch 9 mu var0-0.160443
compute loss for weight  0.226657  0.226662 result 0.537245
   --dy = 0.081921 dy_ref = 0.081921
 training batch 10 mu var0-0.160443
compute loss for weight  0.169433  0.169423 result 0.537243
 training batch 11 mu var0-0.160443
compute loss for weight  0.169413  0.169423 result 0.537247
 training batch 12 mu var0-0.160443
compute loss for weight  0.169428  0.169423 result 0.537244
 training batch 13 mu var0-0.160443
compute loss for weight  0.169418  0.169423 result 0.537246
   --dy = -0.180258 dy_ref = -0.180258
 training batch 14 mu var0-0.160443
compute loss for weight  -1.44076  -1.44077 result 0.537245
 training batch 15 mu var0-0.160443
compute loss for weight  -1.44078  -1.44077 result 0.537245
 training batch 16 mu var0-0.160443
compute loss for weight  -1.44076  -1.44077 result 0.537245
 training batch 17 mu var0-0.160443
compute loss for weight  -1.44077  -1.44077 result 0.537245
   --dy = 0.0282879 dy_ref = 0.0282879
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      0.497      0.5775 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var0-0.160443
compute loss for weight  1.00001  1 result 0.53725
 training batch 19 mu var0-0.160443
compute loss for weight  0.99999  1 result 0.53724
 training batch 20 mu var0-0.160443
compute loss for weight  1.00001  1 result 0.537248
 training batch 21 mu var0-0.160443
compute loss for weight  0.999995  1 result 0.537243
   --dy = 0.496959 dy_ref = 0.496959
 training batch 22 mu var0-0.160443
compute loss for weight  1.00001  1 result 0.537251
 training batch 23 mu var0-0.160443
compute loss for weight  0.99999  1 result 0.537239
 training batch 24 mu var0-0.160443
compute loss for weight  1.00001  1 result 0.537248
 training batch 25 mu var0-0.160443
compute loss for weight  0.999995  1 result 0.537242
   --dy = 0.577531 dy_ref = 0.577531
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 | -1.388e-17  -3.469e-17 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var0-0.160443
compute loss for weight  1e-05  0 result 0.537245
 training batch 27 mu var0-0.160443
compute loss for weight  -1e-05  0 result 0.537245
 training batch 28 mu var0-0.160443
compute loss for weight  5e-06  0 result 0.537245
 training batch 29 mu var0-0.160443
compute loss for weight  -5e-06  0 result 0.537245
   --dy = -1.85037e-12 dy_ref = -1.38778e-17
 training batch 30 mu var0-0.160443
compute loss for weight  1e-05  0 result 0.537245
 training batch 31 mu var0-0.160443
compute loss for weight  -1e-05  0 result 0.537245
 training batch 32 mu var0-0.160443
compute loss for weight  5e-06  0 result 0.537245
 training batch 33 mu var0-0.160443
compute loss for weight  -5e-06  0 result 0.537245
   --dy = 0 dy_ref = -3.46945e-17
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      1.239       1.287 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.4012      0.4486 

 training batch 34 mu var0-0.160443
compute loss for weight  0.401178  0.401168 result 0.537258
 training batch 35 mu var0-0.160443
compute loss for weight  0.401158  0.401168 result 0.537233
 training batch 36 mu var0-0.160443
compute loss for weight  0.401173  0.401168 result 0.537251
 training batch 37 mu var0-0.160443
compute loss for weight  0.401163  0.401168 result 0.537239
   --dy = 1.23878 dy_ref = 1.23878
 training batch 38 mu var0-0.160443
compute loss for weight  0.448606  0.448596 result 0.537258
 training batch 39 mu var0-0.160443
compute loss for weight  0.448586  0.448596 result 0.537232
 training batch 40 mu var0-0.160443
compute loss for weight  0.448601  0.448596 result 0.537252
 training batch 41 mu var0-0.160443
compute loss for weight  0.448591  0.448596 result 0.537239
   --dy = 1.28742 dy_ref = 1.28742
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m3.17502e-10[NON-XML-CHAR-0x1B][39m