Execution Time0.46s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: master-x86_64-mac1015-clang110 (macphsft18.dyndns.cern.ch) on 2019-11-15 00:48:35

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 var00.43783
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.7667        0.82 
   1 |     0.8372      0.4185 
   2 |    -0.3956      -0.305 
   3 |    -0.2142      -1.515 
   4 |     0.5099      0.2952 
   5 |      1.555       1.269 
   6 |     0.2842      0.5231 
   7 |    -0.2722      0.3924 
   8 |      1.417       2.241 
   9 |    -0.1091     -0.3155 

output BN 
output DL feature 0 mean 0.43783	output DL std 0.70119
output DL feature 1 mean 0.382286	output DL std 1.00114
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.4943      0.4608 
   1 |     0.6003     0.03809 
   2 |     -1.253     -0.7236 
   3 |    -0.9802      -1.998 
   4 |     0.1083    -0.09173 
   5 |      1.679      0.9334 
   6 |    -0.2309      0.1483 
   7 |     -1.067     0.01069 
   8 |      1.471       1.957 
   9 |    -0.8221     -0.7347 

output BN feature 0 mean -1.11022e-17	output BN std 1.05397
output BN feature 1 mean 2.22045e-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 |    0.02957     -0.3415     -0.2301     -0.2479 
   1 |  -0.003562      0.3036      0.2212      0.2372 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |     0.8873     0.01632     -0.1359      0.2099 
   1 |     0.8285       0.397      0.2292      0.4744 

 training batch 2 mu var00.437833
compute loss for weight  0.88731  0.8873 result 2.04654
 training batch 3 mu var00.43783
compute loss for weight  0.88729  0.8873 result 2.04654
 training batch 4 mu var00.437831
compute loss for weight  0.887305  0.8873 result 2.04654
 training batch 5 mu var00.43783
compute loss for weight  0.887295  0.8873 result 2.04654
   --dy = 0.0295714 dy_ref = 0.0295714
 training batch 6 mu var00.43783
compute loss for weight  0.0163323  0.0163223 result 2.04653
 training batch 7 mu var00.43783
compute loss for weight  0.0163123  0.0163223 result 2.04654
 training batch 8 mu var00.43783
compute loss for weight  0.0163273  0.0163223 result 2.04654
 training batch 9 mu var00.43783
compute loss for weight  0.0163173  0.0163223 result 2.04654
   --dy = -0.341492 dy_ref = -0.341492
 training batch 10 mu var00.437831
compute loss for weight  -0.135909  -0.135919 result 2.04653
 training batch 11 mu var00.43783
compute loss for weight  -0.135929  -0.135919 result 2.04654
 training batch 12 mu var00.437831
compute loss for weight  -0.135914  -0.135919 result 2.04654
 training batch 13 mu var00.43783
compute loss for weight  -0.135924  -0.135919 result 2.04654
   --dy = -0.230108 dy_ref = -0.230108
 training batch 14 mu var00.43783
compute loss for weight  0.209929  0.209919 result 2.04653
 training batch 15 mu var00.43783
compute loss for weight  0.209909  0.209919 result 2.04654
 training batch 16 mu var00.43783
compute loss for weight  0.209924  0.209919 result 2.04654
 training batch 17 mu var00.43783
compute loss for weight  0.209914  0.209919 result 2.04654
   --dy = -0.247862 dy_ref = -0.247862
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.4002       4.493 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var00.43783
compute loss for weight  1.00001  1 result 2.04653
 training batch 19 mu var00.43783
compute loss for weight  0.99999  1 result 2.04654
 training batch 20 mu var00.43783
compute loss for weight  1.00001  1 result 2.04654
 training batch 21 mu var00.43783
compute loss for weight  0.999995  1 result 2.04654
   --dy = -0.400243 dy_ref = -0.400243
 training batch 22 mu var00.43783
compute loss for weight  1.00001  1 result 2.04658
 training batch 23 mu var00.43783
compute loss for weight  0.99999  1 result 2.04649
 training batch 24 mu var00.43783
compute loss for weight  1.00001  1 result 2.04656
 training batch 25 mu var00.43783
compute loss for weight  0.999995  1 result 2.04651
   --dy = 4.49332 dy_ref = 4.49332
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |  6.939e-18   5.551e-17 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var00.43783
compute loss for weight  1e-05  0 result 2.04654
 training batch 27 mu var00.43783
compute loss for weight  -1e-05  0 result 2.04654
 training batch 28 mu var00.43783
compute loss for weight  5e-06  0 result 2.04654
 training batch 29 mu var00.43783
compute loss for weight  -5e-06  0 result 2.04654
   --dy = -5.92119e-11 dy_ref = 6.93889e-18
 training batch 30 mu var00.43783
compute loss for weight  1e-05  0 result 2.04654
 training batch 31 mu var00.43783
compute loss for weight  -1e-05  0 result 2.04654
 training batch 32 mu var00.43783
compute loss for weight  5e-06  0 result 2.04654
 training batch 33 mu var00.43783
compute loss for weight  -5e-06  0 result 2.04654
   --dy = -7.40149e-12 dy_ref = 5.55112e-17
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      2.189       2.853 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.1829       1.575 

 training batch 34 mu var00.43783
compute loss for weight  -0.18285  -0.18286 result 2.04656
 training batch 35 mu var00.43783
compute loss for weight  -0.18287  -0.18286 result 2.04652
 training batch 36 mu var00.43783
compute loss for weight  -0.182855  -0.18286 result 2.04655
 training batch 37 mu var00.43783
compute loss for weight  -0.182865  -0.18286 result 2.04653
   --dy = 2.1888 dy_ref = 2.1888
 training batch 38 mu var00.43783
compute loss for weight  1.57497  1.57496 result 2.04657
 training batch 39 mu var00.43783
compute loss for weight  1.57495  1.57496 result 2.04651
 training batch 40 mu var00.43783
compute loss for weight  1.57496  1.57496 result 2.04655
 training batch 41 mu var00.43783
compute loss for weight  1.57495  1.57496 result 2.04652
   --dy = 2.85298 dy_ref = 2.85298
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m1.36899e-10[NON-XML-CHAR-0x1B][39m