Execution Time0.56s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: PR-420-x86_64-mac1014-clang100-opt (macitois21.cern.ch) on 2019-11-14 09:35:40

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 var00.165811
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.1592     -0.8288 
   1 |      1.968     -0.2674 
   2 |     -1.426     -0.8195 
   3 |      2.878     -0.4826 
   4 |     0.2775     -0.8371 
   5 |     0.6694      -1.806 
   6 |     -0.207      0.7988 
   7 |      -1.21      0.2563 
   8 |     -1.627     -0.7314 
   9 |     0.4932     0.07248 

output BN 
output DL feature 0 mean 0.165811	output DL std 1.44774
output DL feature 1 mean -0.464517	output DL std 0.723542
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.2367     -0.5307 
   1 |      1.312      0.2872 
   2 |     -1.159      -0.517 
   3 |      1.975    -0.02636 
   4 |    0.08135     -0.5427 
   5 |     0.3667      -1.954 
   6 |    -0.2715        1.84 
   7 |     -1.002        1.05 
   8 |     -1.305     -0.3887 
   9 |     0.2384      0.7822 

output BN feature 0 mean -4.71845e-17	output BN std 1.05406
output BN feature 1 mean 9.99201e-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 |      -0.26      0.1737     -0.2393      0.1665 
   1 |    0.03891     -0.4678     -0.5284     -0.3511 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |     0.7161     -0.5926      -1.462     -0.3651 
   1 |    -0.5546      0.1477     -0.2537      0.4075 

 training batch 2 mu var00.165814
compute loss for weight  0.716113  0.716103 result 2.13488
 training batch 3 mu var00.165811
compute loss for weight  0.716093  0.716103 result 2.13488
 training batch 4 mu var00.165811
compute loss for weight  0.716108  0.716103 result 2.13488
 training batch 5 mu var00.165811
compute loss for weight  0.716098  0.716103 result 2.13488
   --dy = -0.259959 dy_ref = -0.259959
 training batch 6 mu var00.16581
compute loss for weight  -0.592613  -0.592623 result 2.13488
 training batch 7 mu var00.165811
compute loss for weight  -0.592633  -0.592623 result 2.13488
 training batch 8 mu var00.165811
compute loss for weight  -0.592618  -0.592623 result 2.13488
 training batch 9 mu var00.165811
compute loss for weight  -0.592628  -0.592623 result 2.13488
   --dy = 0.173737 dy_ref = 0.173737
 training batch 10 mu var00.165811
compute loss for weight  -1.46226  -1.46227 result 2.13488
 training batch 11 mu var00.165811
compute loss for weight  -1.46228  -1.46227 result 2.13488
 training batch 12 mu var00.165811
compute loss for weight  -1.46226  -1.46227 result 2.13488
 training batch 13 mu var00.165811
compute loss for weight  -1.46227  -1.46227 result 2.13488
   --dy = -0.239289 dy_ref = -0.239289
 training batch 14 mu var00.165811
compute loss for weight  -0.365111  -0.365121 result 2.13488
 training batch 15 mu var00.165811
compute loss for weight  -0.365131  -0.365121 result 2.13488
 training batch 16 mu var00.165811
compute loss for weight  -0.365116  -0.365121 result 2.13488
 training batch 17 mu var00.165811
compute loss for weight  -0.365126  -0.365121 result 2.13488
   --dy = 0.166474 dy_ref = 0.166474
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    0.05027       4.219 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var00.165811
compute loss for weight  1.00001  1 result 2.13488
 training batch 19 mu var00.165811
compute loss for weight  0.99999  1 result 2.13488
 training batch 20 mu var00.165811
compute loss for weight  1.00001  1 result 2.13488
 training batch 21 mu var00.165811
compute loss for weight  0.999995  1 result 2.13488
   --dy = 0.0502693 dy_ref = 0.0502693
 training batch 22 mu var00.165811
compute loss for weight  1.00001  1 result 2.13492
 training batch 23 mu var00.165811
compute loss for weight  0.99999  1 result 2.13484
 training batch 24 mu var00.165811
compute loss for weight  1.00001  1 result 2.1349
 training batch 25 mu var00.165811
compute loss for weight  0.999995  1 result 2.13486
   --dy = 4.21949 dy_ref = 4.21949
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |  7.633e-17   6.661e-16 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var00.165811
compute loss for weight  1e-05  0 result 2.13488
 training batch 27 mu var00.165811
compute loss for weight  -1e-05  0 result 2.13488
 training batch 28 mu var00.165811
compute loss for weight  5e-06  0 result 2.13488
 training batch 29 mu var00.165811
compute loss for weight  -5e-06  0 result 2.13488
   --dy = -7.40149e-12 dy_ref = 7.63278e-17
 training batch 30 mu var00.165811
compute loss for weight  1e-05  0 result 2.13488
 training batch 31 mu var00.165811
compute loss for weight  -1e-05  0 result 2.13488
 training batch 32 mu var00.165811
compute loss for weight  5e-06  0 result 2.13488
 training batch 33 mu var00.165811
compute loss for weight  -5e-06  0 result 2.13488
   --dy = 0 dy_ref = 6.66134e-16
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.2448       2.893 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.2054       1.459 

 training batch 34 mu var00.165811
compute loss for weight  0.205373  0.205363 result 2.13488
 training batch 35 mu var00.165811
compute loss for weight  0.205353  0.205363 result 2.13488
 training batch 36 mu var00.165811
compute loss for weight  0.205368  0.205363 result 2.13488
 training batch 37 mu var00.165811
compute loss for weight  0.205358  0.205363 result 2.13488
   --dy = 0.244783 dy_ref = 0.244783
 training batch 38 mu var00.165811
compute loss for weight  1.45851  1.4585 result 2.13491
 training batch 39 mu var00.165811
compute loss for weight  1.45849  1.4585 result 2.13485
 training batch 40 mu var00.165811
compute loss for weight  1.45851  1.4585 result 2.1349
 training batch 41 mu var00.165811
compute loss for weight  1.4585  1.4585 result 2.13487
   --dy = 2.89303 dy_ref = 2.89303
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m3.82366e-10[NON-XML-CHAR-0x1B][39m