Execution Time0.05s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: master-x86_64-fedora30-gcc9 (root-fedora30-1.cern.ch) on 2019-11-14 01:03:18

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 var00.389622
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.6711      -0.331 
   1 |    -0.6003      -1.421 
   2 |      1.279      0.7771 
   3 |       1.79     -0.2549 
   4 |      1.132     -0.1593 
   5 |      1.712     -0.9003 
   6 |     0.6738      0.8815 
   7 |     -2.956     -0.9715 
   8 |   -0.03014     -0.3345 
   9 |     0.2236     0.03178 

output BN 
output DL feature 0 mean 0.389622	output DL std 1.39792
output DL feature 1 mean -0.268182	output DL std 0.726294
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.2123    -0.09115 
   1 |    -0.7464      -1.673 
   2 |     0.6708       1.517 
   3 |      1.056     0.01925 
   4 |     0.5597       0.158 
   5 |     0.9973     -0.9173 
   6 |     0.2143       1.668 
   7 |     -2.522      -1.021 
   8 |    -0.3165    -0.09621 
   9 |    -0.1252      0.4353 

output BN feature 0 mean -6.38378e-17	output BN std 1.05406
output BN feature 1 mean -1.11022e-17	output BN std 1.05398
Testing weight gradients   for    layer 0
weight gradient for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |      2.869     -0.8523     -0.8255      -1.995 
   1 |      2.212       14.55      -4.929      -1.232 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |    -0.3459       -1.53      0.4487    -0.02993 
   1 |      -1.05     -0.5493      0.6249      0.3143 

 training batch 2 mu var00.389625
compute loss for weight  -0.345904  -0.345914 result 5.17049
 training batch 3 mu var00.389622
compute loss for weight  -0.345924  -0.345914 result 5.17043
 training batch 4 mu var00.389623
compute loss for weight  -0.345909  -0.345914 result 5.17048
 training batch 5 mu var00.389622
compute loss for weight  -0.345919  -0.345914 result 5.17045
   --dy = 2.86932 dy_ref = 2.86932
 training batch 6 mu var00.389622
compute loss for weight  -1.52957  -1.52958 result 5.17045
 training batch 7 mu var00.389622
compute loss for weight  -1.52959  -1.52958 result 5.17047
 training batch 8 mu var00.389622
compute loss for weight  -1.52957  -1.52958 result 5.17046
 training batch 9 mu var00.389622
compute loss for weight  -1.52958  -1.52958 result 5.17047
   --dy = -0.852271 dy_ref = -0.852271
 training batch 10 mu var00.389622
compute loss for weight  0.448679  0.448669 result 5.17045
 training batch 11 mu var00.389622
compute loss for weight  0.448659  0.448669 result 5.17047
 training batch 12 mu var00.389622
compute loss for weight  0.448674  0.448669 result 5.17046
 training batch 13 mu var00.389622
compute loss for weight  0.448664  0.448669 result 5.17047
   --dy = -0.825491 dy_ref = -0.825491
 training batch 14 mu var00.389622
compute loss for weight  -0.0299166  -0.0299266 result 5.17044
 training batch 15 mu var00.389622
compute loss for weight  -0.0299366  -0.0299266 result 5.17048
 training batch 16 mu var00.389622
compute loss for weight  -0.0299216  -0.0299266 result 5.17045
 training batch 17 mu var00.389622
compute loss for weight  -0.0299316  -0.0299266 result 5.17047
   --dy = -1.99491 dy_ref = -1.99491
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      7.164       3.177 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var00.389622
compute loss for weight  1.00001  1 result 5.17053
 training batch 19 mu var00.389622
compute loss for weight  0.99999  1 result 5.17039
 training batch 20 mu var00.389622
compute loss for weight  1.00001  1 result 5.1705
 training batch 21 mu var00.389622
compute loss for weight  0.999995  1 result 5.17043
   --dy = 7.16366 dy_ref = 7.16366
 training batch 22 mu var00.389622
compute loss for weight  1.00001  1 result 5.17049
 training batch 23 mu var00.389622
compute loss for weight  0.99999  1 result 5.17043
 training batch 24 mu var00.389622
compute loss for weight  1.00001  1 result 5.17048
 training batch 25 mu var00.389622
compute loss for weight  0.999995  1 result 5.17045
   --dy = 3.17727 dy_ref = 3.17727
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 | -4.441e-16   9.437e-16 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var00.389622
compute loss for weight  1e-05  0 result 5.17046
 training batch 27 mu var00.389622
compute loss for weight  -1e-05  0 result 5.17046
 training batch 28 mu var00.389622
compute loss for weight  5e-06  0 result 5.17046
 training batch 29 mu var00.389622
compute loss for weight  -5e-06  0 result 5.17046
   --dy = 1.4803e-11 dy_ref = -4.44089e-16
 training batch 30 mu var00.389622
compute loss for weight  1e-05  0 result 5.17046
 training batch 31 mu var00.389622
compute loss for weight  -1e-05  0 result 5.17046
 training batch 32 mu var00.389622
compute loss for weight  5e-06  0 result 5.17046
 training batch 33 mu var00.389622
compute loss for weight  -5e-06  0 result 5.17046
   --dy = -1.4803e-11 dy_ref = 9.4369e-16
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      3.059        -1.7 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      2.342      -1.869 

 training batch 34 mu var00.389622
compute loss for weight  2.34214  2.34213 result 5.17049
 training batch 35 mu var00.389622
compute loss for weight  2.34212  2.34213 result 5.17043
 training batch 36 mu var00.389622
compute loss for weight  2.34214  2.34213 result 5.17048
 training batch 37 mu var00.389622
compute loss for weight  2.34213  2.34213 result 5.17045
   --dy = 3.0586 dy_ref = 3.0586
 training batch 38 mu var00.389622
compute loss for weight  -1.8689  -1.86891 result 5.17045
 training batch 39 mu var00.389622
compute loss for weight  -1.86892  -1.86891 result 5.17048
 training batch 40 mu var00.389622
compute loss for weight  -1.86891  -1.86891 result 5.17045
 training batch 41 mu var00.389622
compute loss for weight  -1.86892  -1.86891 result 5.17047
   --dy = -1.70007 dy_ref = -1.70007
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m2.31956e-10[NON-XML-CHAR-0x1B][39m