Execution Time0.77s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: PR-420-x86_64-fedora30-gcc9-opt (root-fedora30-1.cern.ch) on 2019-11-14 09:33:27

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 var00.307695
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.7809     -0.3114 
   1 |     -1.159      0.7141 
   2 |      1.717      -1.054 
   3 |    -0.5452     -0.9147 
   4 |      0.744     -0.6113 
   5 |      1.266     -0.7649 
   6 |   -0.05667     -0.3184 
   7 |    -0.6722        1.66 
   8 |      1.211      0.1342 
   9 |    -0.2085     -0.1116 

output BN 
output DL feature 0 mean 0.307695	output DL std 0.964401
output DL feature 1 mean -0.157817	output DL std 0.826833
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.5172     -0.1958 
   1 |     -1.603       1.111 
   2 |       1.54      -1.143 
   3 |    -0.9322     -0.9648 
   4 |     0.4768     -0.5781 
   5 |      1.047     -0.7739 
   6 |    -0.3982     -0.2047 
   7 |     -1.071       2.317 
   8 |     0.9874      0.3723 
   9 |    -0.5641     0.05893 

output BN feature 0 mean 1.22125e-16	output BN std 1.05403
output BN feature 1 mean -3.19189e-17	output BN std 1.05401
Testing weight gradients   for    layer 0
weight gradient for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |    -0.1288     -0.8874     -0.3459     -0.3029 
   1 |    -0.1661      -2.326     -0.5168     -0.7452 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |      -0.24     -0.3774       1.094    -0.04229 
   1 |     0.4365      0.8977     -0.4546     0.05748 

 training batch 2 mu var00.307698
compute loss for weight  -0.239948  -0.239958 result 2.0871
 training batch 3 mu var00.307695
compute loss for weight  -0.239968  -0.239958 result 2.0871
 training batch 4 mu var00.307696
compute loss for weight  -0.239953  -0.239958 result 2.0871
 training batch 5 mu var00.307695
compute loss for weight  -0.239963  -0.239958 result 2.0871
   --dy = -0.128833 dy_ref = -0.128833
 training batch 6 mu var00.307694
compute loss for weight  -0.377382  -0.377392 result 2.08709
 training batch 7 mu var00.307695
compute loss for weight  -0.377402  -0.377392 result 2.08711
 training batch 8 mu var00.307695
compute loss for weight  -0.377387  -0.377392 result 2.08709
 training batch 9 mu var00.307695
compute loss for weight  -0.377397  -0.377392 result 2.0871
   --dy = -0.887436 dy_ref = -0.887436
 training batch 10 mu var00.307695
compute loss for weight  1.09399  1.09398 result 2.0871
 training batch 11 mu var00.307695
compute loss for weight  1.09397  1.09398 result 2.0871
 training batch 12 mu var00.307695
compute loss for weight  1.09399  1.09398 result 2.0871
 training batch 13 mu var00.307695
compute loss for weight  1.09398  1.09398 result 2.0871
   --dy = -0.345923 dy_ref = -0.345923
 training batch 14 mu var00.307695
compute loss for weight  -0.0422756  -0.0422856 result 2.0871
 training batch 15 mu var00.307695
compute loss for weight  -0.0422956  -0.0422856 result 2.0871
 training batch 16 mu var00.307695
compute loss for weight  -0.0422806  -0.0422856 result 2.0871
 training batch 17 mu var00.307695
compute loss for weight  -0.0422906  -0.0422856 result 2.0871
   --dy = -0.302916 dy_ref = -0.302916
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      1.688       2.486 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var00.307695
compute loss for weight  1.00001  1 result 2.08712
 training batch 19 mu var00.307695
compute loss for weight  0.99999  1 result 2.08708
 training batch 20 mu var00.307695
compute loss for weight  1.00001  1 result 2.08711
 training batch 21 mu var00.307695
compute loss for weight  0.999995  1 result 2.08709
   --dy = 1.68842 dy_ref = 1.68842
 training batch 22 mu var00.307695
compute loss for weight  1.00001  1 result 2.08712
 training batch 23 mu var00.307695
compute loss for weight  0.99999  1 result 2.08707
 training batch 24 mu var00.307695
compute loss for weight  1.00001  1 result 2.08711
 training batch 25 mu var00.307695
compute loss for weight  0.999995  1 result 2.08709
   --dy = 2.48577 dy_ref = 2.48577
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |  1.457e-16  -2.637e-16 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var00.307695
compute loss for weight  1e-05  0 result 2.0871
 training batch 27 mu var00.307695
compute loss for weight  -1e-05  0 result 2.0871
 training batch 28 mu var00.307695
compute loss for weight  5e-06  0 result 2.0871
 training batch 29 mu var00.307695
compute loss for weight  -5e-06  0 result 2.0871
   --dy = -5.92119e-11 dy_ref = 1.45717e-16
 training batch 30 mu var00.307695
compute loss for weight  1e-05  0 result 2.0871
 training batch 31 mu var00.307695
compute loss for weight  -1e-05  0 result 2.0871
 training batch 32 mu var00.307695
compute loss for weight  5e-06  0 result 2.0871
 training batch 33 mu var00.307695
compute loss for weight  -5e-06  0 result 2.0871
   --dy = -7.40149e-12 dy_ref = -2.63678e-16
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      2.467      -2.669 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.6845     -0.9313 

 training batch 34 mu var00.307695
compute loss for weight  0.68448  0.68447 result 2.08712
 training batch 35 mu var00.307695
compute loss for weight  0.68446  0.68447 result 2.08707
 training batch 36 mu var00.307695
compute loss for weight  0.684475  0.68447 result 2.08711
 training batch 37 mu var00.307695
compute loss for weight  0.684465  0.68447 result 2.08709
   --dy = 2.46676 dy_ref = 2.46676
 training batch 38 mu var00.307695
compute loss for weight  -0.931257  -0.931267 result 2.08707
 training batch 39 mu var00.307695
compute loss for weight  -0.931277  -0.931267 result 2.08713
 training batch 40 mu var00.307695
compute loss for weight  -0.931262  -0.931267 result 2.08709
 training batch 41 mu var00.307695
compute loss for weight  -0.931272  -0.931267 result 2.08711
   --dy = -2.66924 dy_ref = -2.66924
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m3.03954e-10[NON-XML-CHAR-0x1B][39m