Execution Time0.30s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: master-i686-ubuntu18-gcc7 (sft-ubuntu-1804-i386-2) on 2019-11-13 02:23:28
Repository revision: 30660dce2d9e89e4852dbf83dbd8b2cfcc137eff

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 var00.0680459
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.2807      -1.462 
   1 |    -0.3302       -0.31 
   2 |      0.458       0.178 
   3 |     -1.265       2.216 
   4 |  -0.005854     -0.7069 
   5 |     0.1925      -2.332 
   6 |    -0.4312      -1.401 
   7 |      1.076      0.3367 
   8 |     0.9706      -3.757 
   9 |    -0.2656      0.4662 

output BN 
output DL feature 0 mean 0.0680459	output DL std 0.694867
output DL feature 1 mean -0.677114	output DL std 1.66454
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.3225     -0.4972 
   1 |     -0.604      0.2324 
   2 |     0.5915      0.5415 
   3 |     -2.022       1.832 
   4 |    -0.1121    -0.01885 
   5 |     0.1888      -1.048 
   6 |    -0.7572     -0.4583 
   7 |      1.529       0.642 
   8 |      1.369       -1.95 
   9 |    -0.5061       0.724 

output BN feature 0 mean 3.33067e-17	output BN std 1.05397
output BN feature 1 mean 6.66134e-17	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 |    -0.2272      0.1605     -0.1293     -0.2543 
   1 |    -0.0108      0.1247     0.07223     -0.0197 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |     0.1505      0.5505      0.4718    -0.02698 
   1 |     -1.109     -0.1641     -0.5816     -0.9425 

 training batch 2 mu var00.0680487
compute loss for weight  0.150529  0.150519 result 0.153685
 training batch 3 mu var00.0680459
compute loss for weight  0.150509  0.150519 result 0.15369
 training batch 4 mu var00.0680466
compute loss for weight  0.150524  0.150519 result 0.153686
 training batch 5 mu var00.0680459
compute loss for weight  0.150514  0.150519 result 0.153689
   --dy = -0.227239 dy_ref = -0.227239
 training batch 6 mu var00.0680454
compute loss for weight  0.550521  0.550511 result 0.153689
 training batch 7 mu var00.0680459
compute loss for weight  0.550501  0.550511 result 0.153686
 training batch 8 mu var00.0680457
compute loss for weight  0.550516  0.550511 result 0.153688
 training batch 9 mu var00.0680459
compute loss for weight  0.550506  0.550511 result 0.153687
   --dy = 0.160524 dy_ref = 0.160524
 training batch 10 mu var00.0680462
compute loss for weight  0.471827  0.471817 result 0.153686
 training batch 11 mu var00.0680459
compute loss for weight  0.471807  0.471817 result 0.153689
 training batch 12 mu var00.068046
compute loss for weight  0.471822  0.471817 result 0.153687
 training batch 13 mu var00.0680459
compute loss for weight  0.471812  0.471817 result 0.153688
   --dy = -0.129307 dy_ref = -0.129307
 training batch 14 mu var00.0680458
compute loss for weight  -0.026966  -0.026976 result 0.153685
 training batch 15 mu var00.0680459
compute loss for weight  -0.026986  -0.026976 result 0.15369
 training batch 16 mu var00.0680459
compute loss for weight  -0.026971  -0.026976 result 0.153686
 training batch 17 mu var00.0680459
compute loss for weight  -0.026981  -0.026976 result 0.153689
   --dy = -0.254315 dy_ref = -0.254315
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    0.07651      0.2309 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var00.0680459
compute loss for weight  1.00001  1 result 0.153688
 training batch 19 mu var00.0680459
compute loss for weight  0.99999  1 result 0.153687
 training batch 20 mu var00.0680459
compute loss for weight  1.00001  1 result 0.153688
 training batch 21 mu var00.0680459
compute loss for weight  0.999995  1 result 0.153687
   --dy = 0.0765109 dy_ref = 0.0765109
 training batch 22 mu var00.0680459
compute loss for weight  1.00001  1 result 0.15369
 training batch 23 mu var00.0680459
compute loss for weight  0.99999  1 result 0.153685
 training batch 24 mu var00.0680459
compute loss for weight  1.00001  1 result 0.153689
 training batch 25 mu var00.0680459
compute loss for weight  0.999995  1 result 0.153686
   --dy = 0.230864 dy_ref = 0.230864
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |  1.214e-17   2.776e-17 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var00.0680459
compute loss for weight  1e-05  0 result 0.153687
 training batch 27 mu var00.0680459
compute loss for weight  -1e-05  0 result 0.153687
 training batch 28 mu var00.0680459
compute loss for weight  5e-06  0 result 0.153687
 training batch 29 mu var00.0680459
compute loss for weight  -5e-06  0 result 0.153687
   --dy = 3.23815e-12 dy_ref = 1.21431e-17
 training batch 30 mu var00.0680459
compute loss for weight  1e-05  0 result 0.153687
 training batch 31 mu var00.0680459
compute loss for weight  -1e-05  0 result 0.153687
 training batch 32 mu var00.0680459
compute loss for weight  5e-06  0 result 0.153687
 training batch 33 mu var00.0680459
compute loss for weight  -5e-06  0 result 0.153687
   --dy = 9.25186e-13 dy_ref = 2.77556e-17
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.2122      0.5073 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.3605      0.4551 

 training batch 34 mu var00.0680459
compute loss for weight  0.360532  0.360522 result 0.15369
 training batch 35 mu var00.0680459
compute loss for weight  0.360512  0.360522 result 0.153685
 training batch 36 mu var00.0680459
compute loss for weight  0.360527  0.360522 result 0.153689
 training batch 37 mu var00.0680459
compute loss for weight  0.360517  0.360522 result 0.153686
   --dy = 0.212222 dy_ref = 0.212222
 training batch 38 mu var00.0680459
compute loss for weight  0.455127  0.455117 result 0.153693
 training batch 39 mu var00.0680459
compute loss for weight  0.455107  0.455117 result 0.153682
 training batch 40 mu var00.0680459
compute loss for weight  0.455122  0.455117 result 0.15369
 training batch 41 mu var00.0680459
compute loss for weight  0.455112  0.455117 result 0.153685
   --dy = 0.507263 dy_ref = 0.507263
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m7.5409e-11[NON-XML-CHAR-0x1B][39m