Execution Time0.50s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: master-x86_64-mac1014-clang100 (macitois21.cern.ch) on 2019-11-13 03:18:18

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 var00.13399
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.1115       1.128 
   1 |     0.3832      0.0877 
   2 |   -0.07652        0.87 
   3 |      1.192        1.38 
   4 |     0.3379        1.34 
   5 |     0.5747       2.591 
   6 |    0.09195      0.6389 
   7 |     -1.008      -2.606 
   8 |    -0.4514      0.9598 
   9 |     0.1846      0.1052 

output BN 
output DL feature 0 mean 0.13399	output DL std 0.588277
output DL feature 1 mean 0.649564	output DL std 1.34842
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |   -0.04021      0.3743 
   1 |     0.4465     -0.4392 
   2 |    -0.3771      0.1723 
   3 |      1.896      0.5712 
   4 |     0.3653      0.5398 
   5 |     0.7895       1.518 
   6 |   -0.07531   -0.008327 
   7 |     -2.047      -2.545 
   8 |     -1.049      0.2425 
   9 |    0.09067     -0.4255 

output BN feature 0 mean 3.60822e-17	output BN std 1.05392
output BN feature 1 mean -1.66533e-17	output BN std 1.05406
Testing weight gradients   for    layer 0
weight gradient for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |     0.9002     -0.6029        1.25      0.4565 
   1 |    -0.4405      0.1683      -0.696     -0.2904 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |     0.1077     -0.5144     -0.2853     -0.1098 
   1 |     0.3534      -1.249      0.2994     0.06342 

 training batch 2 mu var00.133993
compute loss for weight  0.107755  0.107745 result 3.116
 training batch 3 mu var00.13399
compute loss for weight  0.107735  0.107745 result 3.11598
 training batch 4 mu var00.133991
compute loss for weight  0.10775  0.107745 result 3.116
 training batch 5 mu var00.13399
compute loss for weight  0.10774  0.107745 result 3.11599
   --dy = 0.900218 dy_ref = 0.900218
 training batch 6 mu var00.13399
compute loss for weight  -0.514397  -0.514407 result 3.11599
 training batch 7 mu var00.13399
compute loss for weight  -0.514417  -0.514407 result 3.116
 training batch 8 mu var00.13399
compute loss for weight  -0.514402  -0.514407 result 3.11599
 training batch 9 mu var00.13399
compute loss for weight  -0.514412  -0.514407 result 3.11599
   --dy = -0.602937 dy_ref = -0.602937
 training batch 10 mu var00.13399
compute loss for weight  -0.28527  -0.28528 result 3.116
 training batch 11 mu var00.13399
compute loss for weight  -0.28529  -0.28528 result 3.11598
 training batch 12 mu var00.13399
compute loss for weight  -0.285275  -0.28528 result 3.116
 training batch 13 mu var00.13399
compute loss for weight  -0.285285  -0.28528 result 3.11599
   --dy = 1.2499 dy_ref = 1.2499
 training batch 14 mu var00.13399
compute loss for weight  -0.109835  -0.109845 result 3.116
 training batch 15 mu var00.13399
compute loss for weight  -0.109855  -0.109845 result 3.11599
 training batch 16 mu var00.13399
compute loss for weight  -0.10984  -0.109845 result 3.11599
 training batch 17 mu var00.13399
compute loss for weight  -0.10985  -0.109845 result 3.11599
   --dy = 0.456461 dy_ref = 0.456461
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      1.359       4.873 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var00.13399
compute loss for weight  1.00001  1 result 3.11601
 training batch 19 mu var00.13399
compute loss for weight  0.99999  1 result 3.11598
 training batch 20 mu var00.13399
compute loss for weight  1.00001  1 result 3.116
 training batch 21 mu var00.13399
compute loss for weight  0.999995  1 result 3.11598
   --dy = 1.35932 dy_ref = 1.35932
 training batch 22 mu var00.13399
compute loss for weight  1.00001  1 result 3.11604
 training batch 23 mu var00.13399
compute loss for weight  0.99999  1 result 3.11594
 training batch 24 mu var00.13399
compute loss for weight  1.00001  1 result 3.11602
 training batch 25 mu var00.13399
compute loss for weight  0.999995  1 result 3.11597
   --dy = 4.87266 dy_ref = 4.87266
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 | -1.388e-17   -2.22e-16 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var00.13399
compute loss for weight  1e-05  0 result 3.11599
 training batch 27 mu var00.13399
compute loss for weight  -1e-05  0 result 3.11599
 training batch 28 mu var00.13399
compute loss for weight  5e-06  0 result 3.11599
 training batch 29 mu var00.13399
compute loss for weight  -5e-06  0 result 3.11599
   --dy = -7.40149e-12 dy_ref = -1.38778e-17
 training batch 30 mu var00.13399
compute loss for weight  1e-05  0 result 3.11599
 training batch 31 mu var00.13399
compute loss for weight  -1e-05  0 result 3.11599
 training batch 32 mu var00.13399
compute loss for weight  5e-06  0 result 3.11599
 training batch 33 mu var00.13399
compute loss for weight  -5e-06  0 result 3.11599
   --dy = 5.92119e-11 dy_ref = -2.22045e-16
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     -2.928       -3.47 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.4643      -1.404 

 training batch 34 mu var00.13399
compute loss for weight  -0.4643  -0.46431 result 3.11596
 training batch 35 mu var00.13399
compute loss for weight  -0.46432  -0.46431 result 3.11602
 training batch 36 mu var00.13399
compute loss for weight  -0.464305  -0.46431 result 3.11598
 training batch 37 mu var00.13399
compute loss for weight  -0.464315  -0.46431 result 3.11601
   --dy = -2.92761 dy_ref = -2.92761
 training batch 38 mu var00.13399
compute loss for weight  -1.40438  -1.40439 result 3.11596
 training batch 39 mu var00.13399
compute loss for weight  -1.4044  -1.40439 result 3.11603
 training batch 40 mu var00.13399
compute loss for weight  -1.40438  -1.40439 result 3.11597
 training batch 41 mu var00.13399
compute loss for weight  -1.40439  -1.40439 result 3.11601
   --dy = -3.4696 dy_ref = -3.4696
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m3.23417e-10[NON-XML-CHAR-0x1B][39m