Execution Time0.05s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: PR-4625-x86_64-ubuntu16-gcc54-opt (sft-ubuntu-1604-4) on 2019-11-15 11:04:18
Repository revision: 46b07050c679b2f888ac203c01208696a9c20397

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 var0-0.762011
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      -1.26     0.02412 
   1 |    -0.7587     -0.9494 
   2 |       -0.9      0.7645 
   3 |     -1.497      -1.277 
   4 |     -1.405      -0.151 
   5 |     -2.981     -0.3949 
   6 |     0.8285     -0.1398 
   7 |       1.21       0.739 
   8 |    -0.8216      0.5856 
   9 |   -0.03652     -0.2262 

output BN 
output DL feature 0 mean -0.762011	output DL std 1.20612
output DL feature 1 mean -0.102523	output DL std 0.679139
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.4351      0.1965 
   1 |   0.002912      -1.314 
   2 |    -0.1206       1.346 
   3 |    -0.6426      -1.823 
   4 |    -0.5615     -0.0753 
   5 |     -1.939     -0.4537 
   6 |       1.39    -0.05787 
   7 |      1.724       1.306 
   8 |   -0.05206       1.068 
   9 |      0.634     -0.1919 

output BN feature 0 mean -9.99201e-17	output BN std 1.05405
output BN feature 1 mean -2.77556e-17	output BN std 1.05397
Testing weight gradients   for    layer 0
weight gradient for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |    0.06541      0.1109      0.2796     0.02772 
   1 |    -0.3028     -0.5141      -1.295     -0.1286 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |    -0.9173      0.6088    -0.08016      0.5392 
   1 |    -0.3687      0.3355      0.6907     0.05647 

 training batch 2 mu var0-0.762008
compute loss for weight  -0.917274  -0.917284 result 0.602728
 training batch 3 mu var0-0.762011
compute loss for weight  -0.917294  -0.917284 result 0.602727
 training batch 4 mu var0-0.762011
compute loss for weight  -0.917279  -0.917284 result 0.602728
 training batch 5 mu var0-0.762011
compute loss for weight  -0.917289  -0.917284 result 0.602727
   --dy = 0.0654078 dy_ref = 0.0654078
 training batch 6 mu var0-0.762012
compute loss for weight  0.60882  0.60881 result 0.602729
 training batch 7 mu var0-0.762011
compute loss for weight  0.6088  0.60881 result 0.602727
 training batch 8 mu var0-0.762011
compute loss for weight  0.608815  0.60881 result 0.602728
 training batch 9 mu var0-0.762011
compute loss for weight  0.608805  0.60881 result 0.602727
   --dy = 0.11088 dy_ref = 0.11088
 training batch 10 mu var0-0.762011
compute loss for weight  -0.0801515  -0.0801615 result 0.602731
 training batch 11 mu var0-0.762011
compute loss for weight  -0.0801715  -0.0801615 result 0.602725
 training batch 12 mu var0-0.762011
compute loss for weight  -0.0801565  -0.0801615 result 0.602729
 training batch 13 mu var0-0.762011
compute loss for weight  -0.0801665  -0.0801615 result 0.602726
   --dy = 0.279574 dy_ref = 0.279574
 training batch 14 mu var0-0.762011
compute loss for weight  0.539206  0.539196 result 0.602728
 training batch 15 mu var0-0.762011
compute loss for weight  0.539186  0.539196 result 0.602728
 training batch 16 mu var0-0.762011
compute loss for weight  0.539201  0.539196 result 0.602728
 training batch 17 mu var0-0.762011
compute loss for weight  0.539191  0.539196 result 0.602728
   --dy = 0.027718 dy_ref = 0.027718
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.5424      0.6631 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var0-0.762011
compute loss for weight  1.00001  1 result 0.602733
 training batch 19 mu var0-0.762011
compute loss for weight  0.99999  1 result 0.602722
 training batch 20 mu var0-0.762011
compute loss for weight  1.00001  1 result 0.60273
 training batch 21 mu var0-0.762011
compute loss for weight  0.999995  1 result 0.602725
   --dy = 0.542404 dy_ref = 0.542404
 training batch 22 mu var0-0.762011
compute loss for weight  1.00001  1 result 0.602734
 training batch 23 mu var0-0.762011
compute loss for weight  0.99999  1 result 0.602721
 training batch 24 mu var0-0.762011
compute loss for weight  1.00001  1 result 0.602731
 training batch 25 mu var0-0.762011
compute loss for weight  0.999995  1 result 0.602724
   --dy = 0.663052 dy_ref = 0.663052
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 | -7.286e-17  -2.776e-17 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var0-0.762011
compute loss for weight  1e-05  0 result 0.602728
 training batch 27 mu var0-0.762011
compute loss for weight  -1e-05  0 result 0.602728
 training batch 28 mu var0-0.762011
compute loss for weight  5e-06  0 result 0.602728
 training batch 29 mu var0-0.762011
compute loss for weight  -5e-06  0 result 0.602728
   --dy = 0 dy_ref = -7.28584e-17
 training batch 30 mu var0-0.762011
compute loss for weight  1e-05  0 result 0.602728
 training batch 31 mu var0-0.762011
compute loss for weight  -1e-05  0 result 0.602728
 training batch 32 mu var0-0.762011
compute loss for weight  5e-06  0 result 0.602728
 training batch 33 mu var0-0.762011
compute loss for weight  -5e-06  0 result 0.602728
   --dy = 1.85037e-12 dy_ref = -2.77556e-17
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |       1.25        1.33 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.4339      0.4986 

 training batch 34 mu var0-0.762011
compute loss for weight  0.433881  0.433871 result 0.60274
 training batch 35 mu var0-0.762011
compute loss for weight  0.433861  0.433871 result 0.602715
 training batch 36 mu var0-0.762011
compute loss for weight  0.433876  0.433871 result 0.602734
 training batch 37 mu var0-0.762011
compute loss for weight  0.433866  0.433871 result 0.602722
   --dy = 1.25015 dy_ref = 1.25015
 training batch 38 mu var0-0.762011
compute loss for weight  0.498622  0.498612 result 0.602741
 training batch 39 mu var0-0.762011
compute loss for weight  0.498602  0.498612 result 0.602714
 training batch 40 mu var0-0.762011
compute loss for weight  0.498617  0.498612 result 0.602734
 training batch 41 mu var0-0.762011
compute loss for weight  0.498607  0.498612 result 0.602721
   --dy = 1.3298 dy_ref = 1.3298
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m5.97004e-10[NON-XML-CHAR-0x1B][39m