Execution Time0.06s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: master-x86_64-centos7-gcc48 (lcgapp-centos7-x86-64-25.cern.ch) on 2019-11-13 01:49:52

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 var00.0301287
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.1961       1.681 
   1 |     -1.118       1.216 
   2 |      2.108     -0.4773 
   3 |  -0.001666      -1.459 
   4 |     0.3684       0.965 
   5 |     0.2825       3.078 
   6 |     -2.033      0.6902 
   7 |      1.068     -0.1061 
   8 |    -0.4182       3.597 
   9 |    -0.1504      -0.401 

output BN 
output DL feature 0 mean 0.0301287	output DL std 1.12607
output DL feature 1 mean 0.878358	output DL std 1.59643
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.1554      0.5302 
   1 |     -1.075      0.2226 
   2 |      1.945     -0.8951 
   3 |   -0.02976      -1.543 
   4 |     0.3166     0.05722 
   5 |     0.2362       1.452 
   6 |     -1.931     -0.1242 
   7 |     0.9718       -0.65 
   8 |    -0.4197       1.795 
   9 |     -0.169     -0.8447 

output BN feature 0 mean 8.32667e-18	output BN std 1.05405
output BN feature 1 mean 0	output BN std 1.05407
Testing weight gradients   for    layer 0
weight gradient for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |    -0.2423  -0.0007136     -0.2284      -0.109 
   1 |     -1.058     0.01853      -0.936     -0.5595 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |    -0.4274      0.2761      0.9133     -0.9685 
   1 |      1.707      0.2787      0.1465       0.589 

 training batch 2 mu var00.0301316
compute loss for weight  -0.427425  -0.427435 result 1.67974
 training batch 3 mu var00.0301287
compute loss for weight  -0.427445  -0.427435 result 1.67975
 training batch 4 mu var00.0301294
compute loss for weight  -0.42743  -0.427435 result 1.67975
 training batch 5 mu var00.0301287
compute loss for weight  -0.42744  -0.427435 result 1.67975
   --dy = -0.242305 dy_ref = -0.242305
 training batch 6 mu var00.0301282
compute loss for weight  0.276082  0.276072 result 1.67975
 training batch 7 mu var00.0301287
compute loss for weight  0.276062  0.276072 result 1.67975
 training batch 8 mu var00.0301286
compute loss for weight  0.276077  0.276072 result 1.67975
 training batch 9 mu var00.0301287
compute loss for weight  0.276067  0.276072 result 1.67975
   --dy = -0.000713633 dy_ref = -0.000713633
 training batch 10 mu var00.030129
compute loss for weight  0.913346  0.913336 result 1.67974
 training batch 11 mu var00.0301287
compute loss for weight  0.913326  0.913336 result 1.67975
 training batch 12 mu var00.0301289
compute loss for weight  0.913341  0.913336 result 1.67975
 training batch 13 mu var00.0301287
compute loss for weight  0.913331  0.913336 result 1.67975
   --dy = -0.228433 dy_ref = -0.228433
 training batch 14 mu var00.0301287
compute loss for weight  -0.968535  -0.968545 result 1.67975
 training batch 15 mu var00.0301287
compute loss for weight  -0.968555  -0.968545 result 1.67975
 training batch 16 mu var00.0301287
compute loss for weight  -0.96854  -0.968545 result 1.67975
 training batch 17 mu var00.0301287
compute loss for weight  -0.96855  -0.968545 result 1.67975
   --dy = -0.108972 dy_ref = -0.108972
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      3.216      0.1434 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var00.0301287
compute loss for weight  1.00001  1 result 1.67978
 training batch 19 mu var00.0301287
compute loss for weight  0.99999  1 result 1.67971
 training batch 20 mu var00.0301287
compute loss for weight  1.00001  1 result 1.67976
 training batch 21 mu var00.0301287
compute loss for weight  0.999995  1 result 1.67973
   --dy = 3.21614 dy_ref = 3.21614
 training batch 22 mu var00.0301287
compute loss for weight  1.00001  1 result 1.67975
 training batch 23 mu var00.0301287
compute loss for weight  0.99999  1 result 1.67975
 training batch 24 mu var00.0301287
compute loss for weight  1.00001  1 result 1.67975
 training batch 25 mu var00.0301287
compute loss for weight  0.999995  1 result 1.67975
   --dy = 0.143351 dy_ref = 0.143351
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |  7.633e-17  -9.541e-18 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var00.0301287
compute loss for weight  1e-05  0 result 1.67975
 training batch 27 mu var00.0301287
compute loss for weight  -1e-05  0 result 1.67975
 training batch 28 mu var00.0301287
compute loss for weight  5e-06  0 result 1.67975
 training batch 29 mu var00.0301287
compute loss for weight  -5e-06  0 result 1.67975
   --dy = -5.18104e-11 dy_ref = 7.63278e-17
 training batch 30 mu var00.0301287
compute loss for weight  1e-05  0 result 1.67975
 training batch 31 mu var00.0301287
compute loss for weight  -1e-05  0 result 1.67975
 training batch 32 mu var00.0301287
compute loss for weight  5e-06  0 result 1.67975
 training batch 33 mu var00.0301287
compute loss for weight  -5e-06  0 result 1.67975
   --dy = -7.03141e-11 dy_ref = -9.54098e-18
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      2.575     -0.9308 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      1.249      -0.154 

 training batch 34 mu var00.0301287
compute loss for weight  1.24911  1.2491 result 1.67977
 training batch 35 mu var00.0301287
compute loss for weight  1.24909  1.2491 result 1.67972
 training batch 36 mu var00.0301287
compute loss for weight  1.2491  1.2491 result 1.67976
 training batch 37 mu var00.0301287
compute loss for weight  1.24909  1.2491 result 1.67973
   --dy = 2.57477 dy_ref = 2.57477
 training batch 38 mu var00.0301287
compute loss for weight  -0.153991  -0.154001 result 1.67974
 training batch 39 mu var00.0301287
compute loss for weight  -0.154011  -0.154001 result 1.67976
 training batch 40 mu var00.0301287
compute loss for weight  -0.153996  -0.154001 result 1.67974
 training batch 41 mu var00.0301287
compute loss for weight  -0.154006  -0.154001 result 1.67975
   --dy = -0.93084 dy_ref = -0.93084
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][33m1.76508e-07[NON-XML-CHAR-0x1B][39m