Execution Time0.62s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: PR-420-x86_64-fedora27-gcc7-opt (sft-fedora-27-2.cern.ch) on 2019-11-14 09:32:55
Repository revision: f65ec4f03d673e83895e80d84577e9c51e85d725

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 var00.889982
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      1.322      0.7581 
   1 |       2.29      0.3633 
   2 |     -1.066     -0.1764 
   3 |       1.51      -1.079 
   4 |      1.221      0.3567 
   5 |      3.272       1.255 
   6 |     0.3175      0.4562 
   7 |     -1.524      0.0982 
   8 |      1.444       1.873 
   9 |     0.1123     -0.2423 

output BN 
output DL feature 0 mean 0.889982	output DL std 1.46096
output DL feature 1 mean 0.366238	output DL std 0.819541
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.3118       0.504 
   1 |       1.01    -0.00376 
   2 |     -1.411     -0.6978 
   3 |     0.4475      -1.858 
   4 |     0.2386    -0.01225 
   5 |      1.719       1.142 
   6 |     -0.413      0.1156 
   7 |     -1.742     -0.3447 
   8 |        0.4       1.938 
   9 |    -0.5611     -0.7827 

output BN feature 0 mean 1.11022e-17	output BN std 1.05407
output BN feature 1 mean -4.44089e-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 |     -1.338       -1.18      -2.165      -1.975 
   1 |     -5.099        3.51      -1.072      -1.305 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |      1.785     -0.5287     -0.8813     0.07222 
   1 |     0.7075      0.2144      0.1993      0.3768 

 training batch 2 mu var00.889985
compute loss for weight  1.78532  1.78531 result 3.14577
 training batch 3 mu var00.889982
compute loss for weight  1.7853  1.78531 result 3.1458
 training batch 4 mu var00.889983
compute loss for weight  1.78532  1.78531 result 3.14578
 training batch 5 mu var00.889982
compute loss for weight  1.78531  1.78531 result 3.14579
   --dy = -1.33813 dy_ref = -1.33813
 training batch 6 mu var00.889982
compute loss for weight  -0.528678  -0.528688 result 3.14577
 training batch 7 mu var00.889982
compute loss for weight  -0.528698  -0.528688 result 3.1458
 training batch 8 mu var00.889982
compute loss for weight  -0.528683  -0.528688 result 3.14578
 training batch 9 mu var00.889982
compute loss for weight  -0.528693  -0.528688 result 3.14579
   --dy = -1.17986 dy_ref = -1.17986
 training batch 10 mu var00.889982
compute loss for weight  -0.881273  -0.881283 result 3.14576
 training batch 11 mu var00.889982
compute loss for weight  -0.881293  -0.881283 result 3.14581
 training batch 12 mu var00.889982
compute loss for weight  -0.881278  -0.881283 result 3.14577
 training batch 13 mu var00.889982
compute loss for weight  -0.881288  -0.881283 result 3.1458
   --dy = -2.16493 dy_ref = -2.16493
 training batch 14 mu var00.889982
compute loss for weight  0.0722343  0.0722243 result 3.14577
 training batch 15 mu var00.889982
compute loss for weight  0.0722143  0.0722243 result 3.1458
 training batch 16 mu var00.889982
compute loss for weight  0.0722293  0.0722243 result 3.14578
 training batch 17 mu var00.889982
compute loss for weight  0.0722193  0.0722243 result 3.14579
   --dy = -1.97462 dy_ref = -1.97462
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      1.998       4.293 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var00.889982
compute loss for weight  1.00001  1 result 3.1458
 training batch 19 mu var00.889982
compute loss for weight  0.99999  1 result 3.14577
 training batch 20 mu var00.889982
compute loss for weight  1.00001  1 result 3.1458
 training batch 21 mu var00.889982
compute loss for weight  0.999995  1 result 3.14578
   --dy = 1.99823 dy_ref = 1.99823
 training batch 22 mu var00.889982
compute loss for weight  1.00001  1 result 3.14583
 training batch 23 mu var00.889982
compute loss for weight  0.99999  1 result 3.14574
 training batch 24 mu var00.889982
compute loss for weight  1.00001  1 result 3.14581
 training batch 25 mu var00.889982
compute loss for weight  0.999995  1 result 3.14576
   --dy = 4.29334 dy_ref = 4.29334
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |  4.441e-16    1.11e-16 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var00.889982
compute loss for weight  1e-05  0 result 3.14579
 training batch 27 mu var00.889982
compute loss for weight  -1e-05  0 result 3.14579
 training batch 28 mu var00.889982
compute loss for weight  5e-06  0 result 3.14579
 training batch 29 mu var00.889982
compute loss for weight  -5e-06  0 result 3.14579
   --dy = 1.62833e-10 dy_ref = 4.44089e-16
 training batch 30 mu var00.889982
compute loss for weight  1e-05  0 result 3.14579
 training batch 31 mu var00.889982
compute loss for weight  -1e-05  0 result 3.14579
 training batch 32 mu var00.889982
compute loss for weight  5e-06  0 result 3.14579
 training batch 33 mu var00.889982
compute loss for weight  -5e-06  0 result 3.14579
   --dy = -7.40149e-12 dy_ref = 1.11022e-16
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      1.406      -2.412 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      1.421       -1.78 

 training batch 34 mu var00.889982
compute loss for weight  1.42113  1.42112 result 3.1458
 training batch 35 mu var00.889982
compute loss for weight  1.42111  1.42112 result 3.14577
 training batch 36 mu var00.889982
compute loss for weight  1.42112  1.42112 result 3.14579
 training batch 37 mu var00.889982
compute loss for weight  1.42111  1.42112 result 3.14578
   --dy = 1.4061 dy_ref = 1.4061
 training batch 38 mu var00.889982
compute loss for weight  -1.77975  -1.77976 result 3.14576
 training batch 39 mu var00.889982
compute loss for weight  -1.77977  -1.77976 result 3.14581
 training batch 40 mu var00.889982
compute loss for weight  -1.77975  -1.77976 result 3.14577
 training batch 41 mu var00.889982
compute loss for weight  -1.77976  -1.77976 result 3.1458
   --dy = -2.41231 dy_ref = -2.41231
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m1.62832e-10[NON-XML-CHAR-0x1B][39m