Execution Time0.19s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: master-x86_64-centos7-gcc62-opt-master (olhswep09.cern.ch) on 2019-11-12 23:13:22
Repository revision: 30660dce2d9e89e4852dbf83dbd8b2cfcc137eff

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 var00.480977
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.9212     -0.2156 
   1 |     -0.249       0.707 
   2 |       1.34      -2.228 
   3 |     0.4867      -1.518 
   4 |     0.9965      -0.707 
   5 |      1.926     -0.7713 
   6 |    -0.5916       2.514 
   7 |    -0.7651      -0.364 
   8 |     0.8249       1.329 
   9 |   -0.07981    -0.06802 

output BN 
output DL feature 0 mean 0.480977	output DL std 0.87901
output DL feature 1 mean -0.132141	output DL std 1.37271
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.5279    -0.06406 
   1 |    -0.8753      0.6444 
   2 |       1.03      -1.609 
   3 |   0.006897      -1.064 
   4 |     0.6182     -0.4414 
   5 |      1.733     -0.4908 
   6 |     -1.286       2.032 
   7 |     -1.494      -0.178 
   8 |     0.4124       1.122 
   9 |    -0.6724     0.04923 

output BN feature 0 mean -8.88178e-17	output BN std 1.05402
output BN feature 1 mean 5.75928e-17	output BN std 1.05406
Testing weight gradients   for    layer 0
weight gradient for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |     0.3329    -0.09951      0.1949      0.6842 
   1 |     0.3457     -0.2803      0.3432     0.08062 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |     0.2708     -0.4276      0.5808     -0.3593 
   1 |      0.351      0.1519      -0.587       1.372 

 training batch 2 mu var00.480979
compute loss for weight  0.27076  0.27075 result 0.536163
 training batch 3 mu var00.480977
compute loss for weight  0.27074  0.27075 result 0.536157
 training batch 4 mu var00.480977
compute loss for weight  0.270755  0.27075 result 0.536162
 training batch 5 mu var00.480977
compute loss for weight  0.270745  0.27075 result 0.536158
   --dy = 0.332856 dy_ref = 0.332856
 training batch 6 mu var00.480976
compute loss for weight  -0.427574  -0.427584 result 0.536159
 training batch 7 mu var00.480977
compute loss for weight  -0.427594  -0.427584 result 0.536161
 training batch 8 mu var00.480976
compute loss for weight  -0.427579  -0.427584 result 0.53616
 training batch 9 mu var00.480977
compute loss for weight  -0.427589  -0.427584 result 0.536161
   --dy = -0.0995078 dy_ref = -0.0995078
 training batch 10 mu var00.480977
compute loss for weight  0.580818  0.580808 result 0.536162
 training batch 11 mu var00.480977
compute loss for weight  0.580798  0.580808 result 0.536158
 training batch 12 mu var00.480977
compute loss for weight  0.580813  0.580808 result 0.536161
 training batch 13 mu var00.480977
compute loss for weight  0.580803  0.580808 result 0.536159
   --dy = 0.194899 dy_ref = 0.194899
 training batch 14 mu var00.480977
compute loss for weight  -0.359335  -0.359345 result 0.536167
 training batch 15 mu var00.480977
compute loss for weight  -0.359355  -0.359345 result 0.536153
 training batch 16 mu var00.480977
compute loss for weight  -0.35934  -0.359345 result 0.536164
 training batch 17 mu var00.480977
compute loss for weight  -0.35935  -0.359345 result 0.536157
   --dy = 0.684204 dy_ref = 0.684204
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    0.01897       1.053 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var00.480977
compute loss for weight  1.00001  1 result 0.53616
 training batch 19 mu var00.480977
compute loss for weight  0.99999  1 result 0.53616
 training batch 20 mu var00.480977
compute loss for weight  1.00001  1 result 0.53616
 training batch 21 mu var00.480977
compute loss for weight  0.999995  1 result 0.53616
   --dy = 0.0189712 dy_ref = 0.0189712
 training batch 22 mu var00.480977
compute loss for weight  1.00001  1 result 0.536171
 training batch 23 mu var00.480977
compute loss for weight  0.99999  1 result 0.53615
 training batch 24 mu var00.480977
compute loss for weight  1.00001  1 result 0.536165
 training batch 25 mu var00.480977
compute loss for weight  0.999995  1 result 0.536155
   --dy = 1.05335 dy_ref = 1.05335
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 | -6.939e-18  -6.245e-17 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var00.480977
compute loss for weight  1e-05  0 result 0.53616
 training batch 27 mu var00.480977
compute loss for weight  -1e-05  0 result 0.53616
 training batch 28 mu var00.480977
compute loss for weight  5e-06  0 result 0.53616
 training batch 29 mu var00.480977
compute loss for weight  -5e-06  0 result 0.53616
   --dy = -1.85037e-12 dy_ref = -6.93889e-18
 training batch 30 mu var00.480977
compute loss for weight  1e-05  0 result 0.53616
 training batch 31 mu var00.480977
compute loss for weight  -1e-05  0 result 0.53616
 training batch 32 mu var00.480977
compute loss for weight  5e-06  0 result 0.53616
 training batch 33 mu var00.480977
compute loss for weight  -5e-06  0 result 0.53616
   --dy = -1.85037e-11 dy_ref = -6.245e-17
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |   -0.03971       -1.22 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.4778     -0.8634 

 training batch 34 mu var00.480977
compute loss for weight  -0.47778  -0.47779 result 0.53616
 training batch 35 mu var00.480977
compute loss for weight  -0.4778  -0.47779 result 0.536161
 training batch 36 mu var00.480977
compute loss for weight  -0.477785  -0.47779 result 0.53616
 training batch 37 mu var00.480977
compute loss for weight  -0.477795  -0.47779 result 0.53616
   --dy = -0.0397061 dy_ref = -0.0397061
 training batch 38 mu var00.480977
compute loss for weight  -0.863404  -0.863414 result 0.536148
 training batch 39 mu var00.480977
compute loss for weight  -0.863424  -0.863414 result 0.536172
 training batch 40 mu var00.480977
compute loss for weight  -0.863409  -0.863414 result 0.536154
 training batch 41 mu var00.480977
compute loss for weight  -0.863419  -0.863414 result 0.536166
   --dy = -1.21998 dy_ref = -1.21998
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m2.8202e-10[NON-XML-CHAR-0x1B][39m