Execution Time0.09s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: master-x86_64-fedora29-gcc8-opt-exp-pyroot (root-fedora29-3.cern.ch) on 2019-11-14 04:55:23

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 var00.261241
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.5472       -0.85 
   1 |     0.3454       -1.09 
   2 |    -0.2828       1.203 
   3 |    -0.9662      0.4895 
   4 |     0.1976     -0.4623 
   5 |     0.8656      -1.659 
   6 |     0.4036      -1.684 
   7 |     0.2079       1.197 
   8 |      1.494      -2.087 
   9 |    -0.1996     0.08524 

output BN 
output DL feature 0 mean 0.261241	output DL std 0.667951
output DL feature 1 mean -0.485815	output DL std 1.19435
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.4512     -0.3214 
   1 |     0.1328     -0.5335 
   2 |    -0.8584       1.491 
   3 |     -1.937      0.8607 
   4 |    -0.1004     0.02078 
   5 |     0.9536      -1.035 
   6 |     0.2246      -1.058 
   7 |   -0.08413       1.485 
   8 |      1.945      -1.413 
   9 |    -0.7272       0.504 

output BN feature 0 mean 4.44089e-17	output BN std 1.05396
output BN feature 1 mean -5.55112e-17	output BN std 1.05405
Testing weight gradients   for    layer 0
weight gradient for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |     0.6791      -2.677      -1.328       1.201 
   1 |      1.748      -2.623     -0.5304       2.394 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |     0.5823      0.2463      0.1059      0.3386 
   1 |     -1.014      0.2901      0.3365     -0.8755 

 training batch 2 mu var00.261244
compute loss for weight  0.582338  0.582328 result 7.22152
 training batch 3 mu var00.261241
compute loss for weight  0.582318  0.582328 result 7.22151
 training batch 4 mu var00.261242
compute loss for weight  0.582333  0.582328 result 7.22152
 training batch 5 mu var00.261241
compute loss for weight  0.582323  0.582328 result 7.22151
   --dy = 0.679138 dy_ref = 0.679138
 training batch 6 mu var00.261241
compute loss for weight  0.24628  0.24627 result 7.22149
 training batch 7 mu var00.261241
compute loss for weight  0.24626  0.24627 result 7.22154
 training batch 8 mu var00.261241
compute loss for weight  0.246275  0.24627 result 7.2215
 training batch 9 mu var00.261241
compute loss for weight  0.246265  0.24627 result 7.22153
   --dy = -2.67689 dy_ref = -2.67689
 training batch 10 mu var00.261241
compute loss for weight  0.105861  0.105851 result 7.2215
 training batch 11 mu var00.261241
compute loss for weight  0.105841  0.105851 result 7.22153
 training batch 12 mu var00.261241
compute loss for weight  0.105856  0.105851 result 7.22151
 training batch 13 mu var00.261241
compute loss for weight  0.105846  0.105851 result 7.22152
   --dy = -1.32795 dy_ref = -1.32795
 training batch 14 mu var00.261241
compute loss for weight  0.338568  0.338558 result 7.22153
 training batch 15 mu var00.261241
compute loss for weight  0.338548  0.338558 result 7.22151
 training batch 16 mu var00.261241
compute loss for weight  0.338563  0.338558 result 7.22152
 training batch 17 mu var00.261241
compute loss for weight  0.338553  0.338558 result 7.22151
   --dy = 1.20057 dy_ref = 1.20057
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      8.599       5.844 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var00.261241
compute loss for weight  1.00001  1 result 7.2216
 training batch 19 mu var00.261241
compute loss for weight  0.99999  1 result 7.22143
 training batch 20 mu var00.261241
compute loss for weight  1.00001  1 result 7.22156
 training batch 21 mu var00.261241
compute loss for weight  0.999995  1 result 7.22147
   --dy = 8.59874 dy_ref = 8.59874
 training batch 22 mu var00.261241
compute loss for weight  1.00001  1 result 7.22158
 training batch 23 mu var00.261241
compute loss for weight  0.99999  1 result 7.22146
 training batch 24 mu var00.261241
compute loss for weight  1.00001  1 result 7.22155
 training batch 25 mu var00.261241
compute loss for weight  0.999995  1 result 7.22149
   --dy = 5.84429 dy_ref = 5.84429
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |          0    1.11e-16 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var00.261241
compute loss for weight  1e-05  0 result 7.22152
 training batch 27 mu var00.261241
compute loss for weight  -1e-05  0 result 7.22152
 training batch 28 mu var00.261241
compute loss for weight  5e-06  0 result 7.22152
 training batch 29 mu var00.261241
compute loss for weight  -5e-06  0 result 7.22152
   --dy = 3.55271e-10 dy_ref = 0
 training batch 30 mu var00.261241
compute loss for weight  1e-05  0 result 7.22152
 training batch 31 mu var00.261241
compute loss for weight  -1e-05  0 result 7.22152
 training batch 32 mu var00.261241
compute loss for weight  5e-06  0 result 7.22152
 training batch 33 mu var00.261241
compute loss for weight  -5e-06  0 result 7.22152
   --dy = 0 dy_ref = 1.11022e-16
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |       5.15      -4.922 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |       1.67      -1.187 

 training batch 34 mu var00.261241
compute loss for weight  1.66961  1.6696 result 7.22157
 training batch 35 mu var00.261241
compute loss for weight  1.66959  1.6696 result 7.22147
 training batch 36 mu var00.261241
compute loss for weight  1.6696  1.6696 result 7.22154
 training batch 37 mu var00.261241
compute loss for weight  1.66959  1.6696 result 7.22149
   --dy = 5.15018 dy_ref = 5.15018
 training batch 38 mu var00.261241
compute loss for weight  -1.18732  -1.18733 result 7.22147
 training batch 39 mu var00.261241
compute loss for weight  -1.18734  -1.18733 result 7.22157
 training batch 40 mu var00.261241
compute loss for weight  -1.18732  -1.18733 result 7.22149
 training batch 41 mu var00.261241
compute loss for weight  -1.18733  -1.18733 result 7.22154
   --dy = -4.92222 dy_ref = -4.92222
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m3.55271e-10[NON-XML-CHAR-0x1B][39m