Execution Time0.10s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: master-x86_64-centos7-gcc48 (olhswep22.cern.ch) on 2019-11-14 01:04:51

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 var00.479591
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.6178      0.5238 
   1 |      1.255      0.6569 
   2 |    -0.6363      -2.078 
   3 |      1.875      -2.268 
   4 |     0.8265     -0.2282 
   5 |      1.848      0.4777 
   6 |     0.6657       3.356 
   7 |     -2.077       -1.07 
   8 |     0.1532       3.012 
   9 |     0.2684     -0.2405 

output BN 
output DL feature 0 mean 0.479591	output DL std 1.17979
output DL feature 1 mean 0.214107	output DL std 1.87167
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.1234      0.1744 
   1 |     0.6927      0.2493 
   2 |     -0.997      -1.291 
   3 |      1.246      -1.398 
   4 |       0.31     -0.2491 
   5 |      1.223      0.1484 
   6 |     0.1662        1.77 
   7 |     -2.284      -0.723 
   8 |    -0.2916       1.575 
   9 |    -0.1887      -0.256 

output BN feature 0 mean 1.66533e-17	output BN std 1.05405
output BN feature 1 mean 1.11022e-17	output BN std 1.05408
Testing weight gradients   for    layer 0
weight gradient for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |      0.334      0.1908      0.2787        1.19 
   1 |    0.09529     -0.6918     -0.2975     -0.6441 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |     0.7391      -0.928     -0.6665     0.09759 
   1 |     0.7229    -0.05925     -0.2161       1.835 

 training batch 2 mu var00.479594
compute loss for weight  0.739068  0.739058 result 1.72633
 training batch 3 mu var00.479591
compute loss for weight  0.739048  0.739058 result 1.72633
 training batch 4 mu var00.479592
compute loss for weight  0.739063  0.739058 result 1.72633
 training batch 5 mu var00.479591
compute loss for weight  0.739053  0.739058 result 1.72633
   --dy = 0.333956 dy_ref = 0.333956
 training batch 6 mu var00.47959
compute loss for weight  -0.928025  -0.928035 result 1.72633
 training batch 7 mu var00.479591
compute loss for weight  -0.928045  -0.928035 result 1.72633
 training batch 8 mu var00.479591
compute loss for weight  -0.92803  -0.928035 result 1.72633
 training batch 9 mu var00.479591
compute loss for weight  -0.92804  -0.928035 result 1.72633
   --dy = 0.190831 dy_ref = 0.190831
 training batch 10 mu var00.479591
compute loss for weight  -0.666483  -0.666493 result 1.72633
 training batch 11 mu var00.479591
compute loss for weight  -0.666503  -0.666493 result 1.72633
 training batch 12 mu var00.479591
compute loss for weight  -0.666488  -0.666493 result 1.72633
 training batch 13 mu var00.479591
compute loss for weight  -0.666498  -0.666493 result 1.72633
   --dy = 0.278681 dy_ref = 0.278681
 training batch 14 mu var00.479591
compute loss for weight  0.0975979  0.0975879 result 1.72634
 training batch 15 mu var00.479591
compute loss for weight  0.0975779  0.0975879 result 1.72632
 training batch 16 mu var00.479591
compute loss for weight  0.0975929  0.0975879 result 1.72634
 training batch 17 mu var00.479591
compute loss for weight  0.0975829  0.0975879 result 1.72632
   --dy = 1.19022 dy_ref = 1.19022
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      1.599       1.854 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var00.479591
compute loss for weight  1.00001  1 result 1.72635
 training batch 19 mu var00.479591
compute loss for weight  0.99999  1 result 1.72631
 training batch 20 mu var00.479591
compute loss for weight  1.00001  1 result 1.72634
 training batch 21 mu var00.479591
compute loss for weight  0.999995  1 result 1.72632
   --dy = 1.59887 dy_ref = 1.59887
 training batch 22 mu var00.479591
compute loss for weight  1.00001  1 result 1.72635
 training batch 23 mu var00.479591
compute loss for weight  0.99999  1 result 1.72631
 training batch 24 mu var00.479591
compute loss for weight  1.00001  1 result 1.72634
 training batch 25 mu var00.479591
compute loss for weight  0.999995  1 result 1.72632
   --dy = 1.85379 dy_ref = 1.85379
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

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

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var00.479591
compute loss for weight  1e-05  0 result 1.72633
 training batch 27 mu var00.479591
compute loss for weight  -1e-05  0 result 1.72633
 training batch 28 mu var00.479591
compute loss for weight  5e-06  0 result 1.72633
 training batch 29 mu var00.479591
compute loss for weight  -5e-06  0 result 1.72633
   --dy = 5.92119e-11 dy_ref = 0
 training batch 30 mu var00.479591
compute loss for weight  1e-05  0 result 1.72633
 training batch 31 mu var00.479591
compute loss for weight  -1e-05  0 result 1.72633
 training batch 32 mu var00.479591
compute loss for weight  5e-06  0 result 1.72633
 training batch 33 mu var00.479591
compute loss for weight  -5e-06  0 result 1.72633
   --dy = 0 dy_ref = 2.77556e-17
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      1.917       2.044 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.8339       0.907 

 training batch 34 mu var00.479591
compute loss for weight  0.83386  0.83385 result 1.72635
 training batch 35 mu var00.479591
compute loss for weight  0.83384  0.83385 result 1.72631
 training batch 36 mu var00.479591
compute loss for weight  0.833855  0.83385 result 1.72634
 training batch 37 mu var00.479591
compute loss for weight  0.833845  0.83385 result 1.72632
   --dy = 1.91745 dy_ref = 1.91745
 training batch 38 mu var00.479591
compute loss for weight  0.90706  0.90705 result 1.72635
 training batch 39 mu var00.479591
compute loss for weight  0.90704  0.90705 result 1.72631
 training batch 40 mu var00.479591
compute loss for weight  0.907055  0.90705 result 1.72634
 training batch 41 mu var00.479591
compute loss for weight  0.907045  0.90705 result 1.72632
   --dy = 2.04376 dy_ref = 2.04376
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m2.259e-10[NON-XML-CHAR-0x1B][39m