Execution Time0.91s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: master-x86_64-mac1014-clang100-dbg (macitois21.cern.ch) on 2019-11-14 11:42:56

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 var00.757339
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |       1.68      -0.285 
   1 |     0.1514     -0.5778 
   2 |    0.03707     -0.7387 
   3 |     -2.721     -0.2246 
   4 |     0.8015     -0.2575 
   5 |      2.611     -0.7255 
   6 |      1.431        2.64 
   7 |    -0.1809      -2.154 
   8 |      4.339      0.2936 
   9 |    -0.5756      0.1362 

output BN 
output DL feature 0 mean 0.757339	output DL std 1.92107
output DL feature 1 mean -0.189318	output DL std 1.19998
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.5062    -0.08406 
   1 |    -0.3324     -0.3412 
   2 |    -0.3952     -0.4825 
   3 |     -1.909    -0.03103 
   4 |    0.02421    -0.05992 
   5 |      1.017      -0.471 
   6 |     0.3696       2.485 
   7 |    -0.5148      -1.726 
   8 |      1.965      0.4242 
   9 |    -0.7314       0.286 

output BN feature 0 mean 3.33067e-17	output BN std 1.05408
output BN feature 1 mean -1.11022e-17	output BN std 1.05405
Testing weight gradients   for    layer 0
weight gradient for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |   -0.03152    -0.06264     -0.0378     0.08377 
   1 |     0.2426      0.5541      0.2926     -0.7863 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |      1.188      0.2668       0.818       1.016 
   1 |    -0.6552     -0.9183    0.005558       1.135 

 training batch 2 mu var00.757342
compute loss for weight  1.18819  1.18818 result 0.567146
 training batch 3 mu var00.757339
compute loss for weight  1.18817  1.18818 result 0.567146
 training batch 4 mu var00.75734
compute loss for weight  1.18819  1.18818 result 0.567146
 training batch 5 mu var00.757339
compute loss for weight  1.18818  1.18818 result 0.567146
   --dy = -0.0315195 dy_ref = -0.0315195
 training batch 6 mu var00.757338
compute loss for weight  0.26676  0.26675 result 0.567145
 training batch 7 mu var00.757339
compute loss for weight  0.26674  0.26675 result 0.567147
 training batch 8 mu var00.757339
compute loss for weight  0.266755  0.26675 result 0.567146
 training batch 9 mu var00.757339
compute loss for weight  0.266745  0.26675 result 0.567146
   --dy = -0.0626386 dy_ref = -0.0626386
 training batch 10 mu var00.757339
compute loss for weight  0.817997  0.817987 result 0.567146
 training batch 11 mu var00.757339
compute loss for weight  0.817977  0.817987 result 0.567147
 training batch 12 mu var00.757339
compute loss for weight  0.817992  0.817987 result 0.567146
 training batch 13 mu var00.757339
compute loss for weight  0.817982  0.817987 result 0.567146
   --dy = -0.0377966 dy_ref = -0.0377966
 training batch 14 mu var00.757339
compute loss for weight  1.01592  1.01591 result 0.567147
 training batch 15 mu var00.757339
compute loss for weight  1.0159  1.01591 result 0.567145
 training batch 16 mu var00.757339
compute loss for weight  1.01592  1.01591 result 0.567147
 training batch 17 mu var00.757339
compute loss for weight  1.01591  1.01591 result 0.567146
   --dy = 0.083773 dy_ref = 0.083773
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.9739      0.1604 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var00.757339
compute loss for weight  1.00001  1 result 0.567156
 training batch 19 mu var00.757339
compute loss for weight  0.99999  1 result 0.567136
 training batch 20 mu var00.757339
compute loss for weight  1.00001  1 result 0.567151
 training batch 21 mu var00.757339
compute loss for weight  0.999995  1 result 0.567141
   --dy = 0.973865 dy_ref = 0.973865
 training batch 22 mu var00.757339
compute loss for weight  1.00001  1 result 0.567148
 training batch 23 mu var00.757339
compute loss for weight  0.99999  1 result 0.567145
 training batch 24 mu var00.757339
compute loss for weight  1.00001  1 result 0.567147
 training batch 25 mu var00.757339
compute loss for weight  0.999995  1 result 0.567145
   --dy = 0.160427 dy_ref = 0.160427
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

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

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var00.757339
compute loss for weight  1e-05  0 result 0.567146
 training batch 27 mu var00.757339
compute loss for weight  -1e-05  0 result 0.567146
 training batch 28 mu var00.757339
compute loss for weight  5e-06  0 result 0.567146
 training batch 29 mu var00.757339
compute loss for weight  -5e-06  0 result 0.567146
   --dy = -1.66533e-11 dy_ref = 2.77556e-17
 training batch 30 mu var00.757339
compute loss for weight  1e-05  0 result 0.567146
 training batch 31 mu var00.757339
compute loss for weight  -1e-05  0 result 0.567146
 training batch 32 mu var00.757339
compute loss for weight  5e-06  0 result 0.567146
 training batch 33 mu var00.757339
compute loss for weight  -5e-06  0 result 0.567146
   --dy = 0 dy_ref = 0
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     -1.446     -0.7398 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.6736     -0.2169 

 training batch 34 mu var00.757339
compute loss for weight  -0.673607  -0.673617 result 0.567132
 training batch 35 mu var00.757339
compute loss for weight  -0.673627  -0.673617 result 0.567161
 training batch 36 mu var00.757339
compute loss for weight  -0.673612  -0.673617 result 0.567139
 training batch 37 mu var00.757339
compute loss for weight  -0.673622  -0.673617 result 0.567153
   --dy = -1.44573 dy_ref = -1.44573
 training batch 38 mu var00.757339
compute loss for weight  -0.216855  -0.216865 result 0.567139
 training batch 39 mu var00.757339
compute loss for weight  -0.216875  -0.216865 result 0.567154
 training batch 40 mu var00.757339
compute loss for weight  -0.21686  -0.216865 result 0.567142
 training batch 41 mu var00.757339
compute loss for weight  -0.21687  -0.216865 result 0.56715
   --dy = -0.739755 dy_ref = -0.739755
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m4.07319e-10[NON-XML-CHAR-0x1B][39m