Execution Time0.14s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: master-x86_64-centos7-gcc62-opt-master (olsnba08.cern.ch) on 2019-11-14 11:14:29
Repository revision: 14de58de35eff907054671888ccc2de0f7f27e77

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 var0-0.211372
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.2608      0.7659 
   1 |      -0.79       -0.31 
   2 |     0.6109      0.8608 
   3 |     -0.629      -1.504 
   4 |    -0.2684      0.3721 
   5 |    -0.7674       1.095 
   6 |    -0.3843     -0.6427 
   7 |     0.7006       1.086 
   8 |    -0.2288       1.745 
   9 |   -0.09652      -0.379 

output BN 
output DL feature 0 mean -0.211372	output DL std 0.513733
output DL feature 1 mean 0.308823	output DL std 0.991626
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.1015      0.4859 
   1 |     -1.187     -0.6577 
   2 |      1.687      0.5867 
   3 |    -0.8567      -1.927 
   4 |     -0.117      0.0673 
   5 |     -1.141      0.8351 
   6 |    -0.3547      -1.011 
   7 |      1.871      0.8263 
   8 |   -0.03575       1.526 
   9 |     0.2356     -0.7311 

output BN feature 0 mean -2.77556e-18	output BN std 1.05387
output BN feature 1 mean -3.33067e-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 |    -0.3132     -0.0683     -0.3458     -0.1077 
   1 |     0.1076       -0.17    -0.08299      0.0362 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |     -0.485      0.2633      0.4275     -0.1302 
   1 |     0.4483      0.5782      0.7251    -0.08733 

 training batch 2 mu var0-0.211369
compute loss for weight  -0.484961  -0.484971 result 0.16119
 training batch 3 mu var0-0.211372
compute loss for weight  -0.484981  -0.484971 result 0.161196
 training batch 4 mu var0-0.211371
compute loss for weight  -0.484966  -0.484971 result 0.161192
 training batch 5 mu var0-0.211372
compute loss for weight  -0.484976  -0.484971 result 0.161195
   --dy = -0.313182 dy_ref = -0.313182
 training batch 6 mu var0-0.211372
compute loss for weight  0.26332  0.26331 result 0.161192
 training batch 7 mu var0-0.211372
compute loss for weight  0.2633  0.26331 result 0.161194
 training batch 8 mu var0-0.211372
compute loss for weight  0.263315  0.26331 result 0.161193
 training batch 9 mu var0-0.211372
compute loss for weight  0.263305  0.26331 result 0.161193
   --dy = -0.0682961 dy_ref = -0.0682961
 training batch 10 mu var0-0.211372
compute loss for weight  0.427476  0.427466 result 0.16119
 training batch 11 mu var0-0.211372
compute loss for weight  0.427456  0.427466 result 0.161197
 training batch 12 mu var0-0.211372
compute loss for weight  0.427471  0.427466 result 0.161191
 training batch 13 mu var0-0.211372
compute loss for weight  0.427461  0.427466 result 0.161195
   --dy = -0.345829 dy_ref = -0.345829
 training batch 14 mu var0-0.211372
compute loss for weight  -0.130161  -0.130171 result 0.161192
 training batch 15 mu var0-0.211372
compute loss for weight  -0.130181  -0.130171 result 0.161194
 training batch 16 mu var0-0.211372
compute loss for weight  -0.130166  -0.130171 result 0.161193
 training batch 17 mu var0-0.211372
compute loss for weight  -0.130176  -0.130171 result 0.161194
   --dy = -0.107713 dy_ref = -0.107713
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      0.219      0.1034 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var0-0.211372
compute loss for weight  1.00001  1 result 0.161195
 training batch 19 mu var0-0.211372
compute loss for weight  0.99999  1 result 0.161191
 training batch 20 mu var0-0.211372
compute loss for weight  1.00001  1 result 0.161194
 training batch 21 mu var0-0.211372
compute loss for weight  0.999995  1 result 0.161192
   --dy = 0.218953 dy_ref = 0.218953
 training batch 22 mu var0-0.211372
compute loss for weight  1.00001  1 result 0.161194
 training batch 23 mu var0-0.211372
compute loss for weight  0.99999  1 result 0.161192
 training batch 24 mu var0-0.211372
compute loss for weight  1.00001  1 result 0.161194
 training batch 25 mu var0-0.211372
compute loss for weight  0.999995  1 result 0.161193
   --dy = 0.103434 dy_ref = 0.103434
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |  3.469e-18   3.469e-18 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var0-0.211372
compute loss for weight  1e-05  0 result 0.161193
 training batch 27 mu var0-0.211372
compute loss for weight  -1e-05  0 result 0.161193
 training batch 28 mu var0-0.211372
compute loss for weight  5e-06  0 result 0.161193
 training batch 29 mu var0-0.211372
compute loss for weight  -5e-06  0 result 0.161193
   --dy = -3.70074e-12 dy_ref = 3.46945e-18
 training batch 30 mu var0-0.211372
compute loss for weight  1e-05  0 result 0.161193
 training batch 31 mu var0-0.211372
compute loss for weight  -1e-05  0 result 0.161193
 training batch 32 mu var0-0.211372
compute loss for weight  5e-06  0 result 0.161193
 training batch 33 mu var0-0.211372
compute loss for weight  -5e-06  0 result 0.161193
   --dy = -3.70074e-12 dy_ref = 3.46945e-18
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.5421     -0.3187 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.4039     -0.3246 

 training batch 34 mu var0-0.211372
compute loss for weight  0.403941  0.403931 result 0.161199
 training batch 35 mu var0-0.211372
compute loss for weight  0.403921  0.403931 result 0.161188
 training batch 36 mu var0-0.211372
compute loss for weight  0.403936  0.403931 result 0.161196
 training batch 37 mu var0-0.211372
compute loss for weight  0.403926  0.403931 result 0.16119
   --dy = 0.542055 dy_ref = 0.542055
 training batch 38 mu var0-0.211372
compute loss for weight  -0.324557  -0.324567 result 0.16119
 training batch 39 mu var0-0.211372
compute loss for weight  -0.324577  -0.324567 result 0.161196
 training batch 40 mu var0-0.211372
compute loss for weight  -0.324562  -0.324567 result 0.161192
 training batch 41 mu var0-0.211372
compute loss for weight  -0.324572  -0.324567 result 0.161195
   --dy = -0.318682 dy_ref = -0.318682
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m1.18094e-10[NON-XML-CHAR-0x1B][39m