Execution Time0.05s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: master-x86_64-fedora27-gcc7 (sft-fedora-27-2.cern.ch) on 2019-11-15 01:15:04
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 var0-0.384393
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.8454     -0.7017 
   1 |     0.6986      -1.949 
   2 |     -1.158       1.326 
   3 |     0.2166     -0.4643 
   4 |    -0.8084      -0.456 
   5 |     -1.526      -1.738 
   6 |    -0.4789    -0.00955 
   7 |      1.249     0.02429 
   8 |     -1.321      -1.046 
   9 |     0.1299   -0.007307 

output BN 
output DL feature 0 mean -0.384393	output DL std 0.921671
output DL feature 1 mean -0.502214	output DL std 0.946683
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.5272     -0.2221 
   1 |      1.238      -1.611 
   2 |    -0.8848       2.035 
   3 |     0.6873      0.0422 
   4 |    -0.4849     0.05144 
   5 |     -1.305      -1.376 
   6 |    -0.1081      0.5485 
   7 |      1.868      0.5862 
   8 |     -1.071     -0.6055 
   9 |     0.5882       0.551 

output BN feature 0 mean 1.11022e-17	output BN std 1.05402
output BN feature 1 mean 6.66134e-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 |      0.528     -0.1552     -0.1346      0.1294 
   1 |      0.951     -0.6263      0.5426      0.2872 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |    0.01801      0.5923     -0.7817     -0.1752 
   1 |     -1.452     -0.1826      0.8532    -0.07635 

 training batch 2 mu var0-0.38439
compute loss for weight  0.0180198  0.0180098 result 0.829623
 training batch 3 mu var0-0.384393
compute loss for weight  0.0179998  0.0180098 result 0.829613
 training batch 4 mu var0-0.384392
compute loss for weight  0.0180148  0.0180098 result 0.829621
 training batch 5 mu var0-0.384393
compute loss for weight  0.0180048  0.0180098 result 0.829616
   --dy = 0.527957 dy_ref = 0.527957
 training batch 6 mu var0-0.384393
compute loss for weight  0.592334  0.592324 result 0.829617
 training batch 7 mu var0-0.384393
compute loss for weight  0.592314  0.592324 result 0.82962
 training batch 8 mu var0-0.384393
compute loss for weight  0.592329  0.592324 result 0.829617
 training batch 9 mu var0-0.384393
compute loss for weight  0.592319  0.592324 result 0.829619
   --dy = -0.155242 dy_ref = -0.155242
 training batch 10 mu var0-0.384393
compute loss for weight  -0.781707  -0.781717 result 0.829617
 training batch 11 mu var0-0.384393
compute loss for weight  -0.781727  -0.781717 result 0.82962
 training batch 12 mu var0-0.384393
compute loss for weight  -0.781712  -0.781717 result 0.829618
 training batch 13 mu var0-0.384393
compute loss for weight  -0.781722  -0.781717 result 0.829619
   --dy = -0.13461 dy_ref = -0.13461
 training batch 14 mu var0-0.384393
compute loss for weight  -0.175225  -0.175235 result 0.829619
 training batch 15 mu var0-0.384393
compute loss for weight  -0.175245  -0.175235 result 0.829617
 training batch 16 mu var0-0.384393
compute loss for weight  -0.17523  -0.175235 result 0.829619
 training batch 17 mu var0-0.384393
compute loss for weight  -0.17524  -0.175235 result 0.829618
   --dy = 0.129376 dy_ref = 0.129376
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      0.848      0.8112 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var0-0.384393
compute loss for weight  1.00001  1 result 0.829627
 training batch 19 mu var0-0.384393
compute loss for weight  0.99999  1 result 0.82961
 training batch 20 mu var0-0.384393
compute loss for weight  1.00001  1 result 0.829622
 training batch 21 mu var0-0.384393
compute loss for weight  0.999995  1 result 0.829614
   --dy = 0.848028 dy_ref = 0.848028
 training batch 22 mu var0-0.384393
compute loss for weight  1.00001  1 result 0.829626
 training batch 23 mu var0-0.384393
compute loss for weight  0.99999  1 result 0.82961
 training batch 24 mu var0-0.384393
compute loss for weight  1.00001  1 result 0.829622
 training batch 25 mu var0-0.384393
compute loss for weight  0.999995  1 result 0.829614
   --dy = 0.811209 dy_ref = 0.811209
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 | -2.689e-17   3.556e-17 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var0-0.384393
compute loss for weight  1e-05  0 result 0.829618
 training batch 27 mu var0-0.384393
compute loss for weight  -1e-05  0 result 0.829618
 training batch 28 mu var0-0.384393
compute loss for weight  5e-06  0 result 0.829618
 training batch 29 mu var0-0.384393
compute loss for weight  -5e-06  0 result 0.829618
   --dy = -1.85037e-11 dy_ref = -2.68882e-17
 training batch 30 mu var0-0.384393
compute loss for weight  1e-05  0 result 0.829618
 training batch 31 mu var0-0.384393
compute loss for weight  -1e-05  0 result 0.829618
 training batch 32 mu var0-0.384393
compute loss for weight  5e-06  0 result 0.829618
 training batch 33 mu var0-0.384393
compute loss for weight  -5e-06  0 result 0.829618
   --dy = 3.70074e-12 dy_ref = 3.55618e-17
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      1.294      -1.265 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.6554     -0.6412 

 training batch 34 mu var0-0.384393
compute loss for weight  0.655366  0.655356 result 0.829631
 training batch 35 mu var0-0.384393
compute loss for weight  0.655346  0.655356 result 0.829605
 training batch 36 mu var0-0.384393
compute loss for weight  0.655361  0.655356 result 0.829625
 training batch 37 mu var0-0.384393
compute loss for weight  0.655351  0.655356 result 0.829612
   --dy = 1.294 dy_ref = 1.294
 training batch 38 mu var0-0.384393
compute loss for weight  -0.641143  -0.641153 result 0.829606
 training batch 39 mu var0-0.384393
compute loss for weight  -0.641163  -0.641153 result 0.829631
 training batch 40 mu var0-0.384393
compute loss for weight  -0.641148  -0.641153 result 0.829612
 training batch 41 mu var0-0.384393
compute loss for weight  -0.641158  -0.641153 result 0.829625
   --dy = -1.26523 dy_ref = -1.26523
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m1.8575e-10[NON-XML-CHAR-0x1B][39m