Execution Time0.12s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: master-x86_64-centos7-gcc48-dbg (olhswep22.cern.ch) on 2019-11-14 11:42:18

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 var00.228215
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.4417      -0.802 
   1 |    -0.3177      0.3204 
   2 |     0.6551     -0.1249 
   3 |     0.4374       1.521 
   4 |     0.5512     -0.3792 
   5 |     0.9453      -1.134 
   6 |     0.3063        -1.3 
   7 |     -1.151      0.6224 
   8 |     0.3911       -2.29 
   9 |    0.02318      0.2821 

output BN 
output DL feature 0 mean 0.228215	output DL std 0.593132
output DL feature 1 mean -0.328373	output DL std 1.09674
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.3794     -0.4552 
   1 |      -0.97      0.6235 
   2 |     0.7585      0.1955 
   3 |     0.3717       1.777 
   4 |     0.5739    -0.04881 
   5 |      1.274     -0.7744 
   6 |     0.1388     -0.9336 
   7 |     -2.452      0.9138 
   8 |     0.2894      -1.885 
   9 |    -0.3643      0.5867 

output BN feature 0 mean -1.60982e-16	output BN std 1.05393
output BN feature 1 mean 1.11022e-17	output BN std 1.05404
Testing weight gradients   for    layer 0
weight gradient for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |    -0.4613     -0.3896     -0.6632     -0.9937 
   1 |     0.3627      -0.615      0.4232      0.0831 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |   -0.06405     -0.5773      0.3273     0.03738 
   1 |    -0.3149      0.1153     -0.5884     -0.7548 

 training batch 2 mu var00.228217
compute loss for weight  -0.0640384  -0.0640484 result 0.553989
 training batch 3 mu var00.228215
compute loss for weight  -0.0640584  -0.0640484 result 0.553998
 training batch 4 mu var00.228215
compute loss for weight  -0.0640434  -0.0640484 result 0.553991
 training batch 5 mu var00.228215
compute loss for weight  -0.0640534  -0.0640484 result 0.553996
   --dy = -0.461335 dy_ref = -0.461335
 training batch 6 mu var00.228214
compute loss for weight  -0.577304  -0.577314 result 0.55399
 training batch 7 mu var00.228215
compute loss for weight  -0.577324  -0.577314 result 0.553998
 training batch 8 mu var00.228214
compute loss for weight  -0.577309  -0.577314 result 0.553992
 training batch 9 mu var00.228215
compute loss for weight  -0.577319  -0.577314 result 0.553996
   --dy = -0.389628 dy_ref = -0.389628
 training batch 10 mu var00.228215
compute loss for weight  0.327271  0.327261 result 0.553987
 training batch 11 mu var00.228215
compute loss for weight  0.327251  0.327261 result 0.554
 training batch 12 mu var00.228215
compute loss for weight  0.327266  0.327261 result 0.55399
 training batch 13 mu var00.228215
compute loss for weight  0.327256  0.327261 result 0.553997
   --dy = -0.663228 dy_ref = -0.663228
 training batch 14 mu var00.228215
compute loss for weight  0.0373912  0.0373812 result 0.553984
 training batch 15 mu var00.228215
compute loss for weight  0.0373712  0.0373812 result 0.554004
 training batch 16 mu var00.228215
compute loss for weight  0.0373862  0.0373812 result 0.553989
 training batch 17 mu var00.228215
compute loss for weight  0.0373762  0.0373812 result 0.553999
   --dy = -0.993724 dy_ref = -0.993724
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.9187      0.1893 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var00.228215
compute loss for weight  1.00001  1 result 0.554003
 training batch 19 mu var00.228215
compute loss for weight  0.99999  1 result 0.553985
 training batch 20 mu var00.228215
compute loss for weight  1.00001  1 result 0.553998
 training batch 21 mu var00.228215
compute loss for weight  0.999995  1 result 0.553989
   --dy = 0.918656 dy_ref = 0.918656
 training batch 22 mu var00.228215
compute loss for weight  1.00001  1 result 0.553996
 training batch 23 mu var00.228215
compute loss for weight  0.99999  1 result 0.553992
 training batch 24 mu var00.228215
compute loss for weight  1.00001  1 result 0.553995
 training batch 25 mu var00.228215
compute loss for weight  0.999995  1 result 0.553993
   --dy = 0.189331 dy_ref = 0.189331
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 | -2.084e-16   -1.05e-16 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var00.228215
compute loss for weight  1e-05  0 result 0.553994
 training batch 27 mu var00.228215
compute loss for weight  -1e-05  0 result 0.553994
 training batch 28 mu var00.228215
compute loss for weight  5e-06  0 result 0.553994
 training batch 29 mu var00.228215
compute loss for weight  -5e-06  0 result 0.553994
   --dy = 0 dy_ref = -2.08384e-16
 training batch 30 mu var00.228215
compute loss for weight  1e-05  0 result 0.553994
 training batch 31 mu var00.228215
compute loss for weight  -1e-05  0 result 0.553994
 training batch 32 mu var00.228215
compute loss for weight  5e-06  0 result 0.553994
 training batch 33 mu var00.228215
compute loss for weight  -5e-06  0 result 0.553994
   --dy = 2.77556e-11 dy_ref = -1.04951e-16
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      1.161      0.3707 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      0.791      0.5107 

 training batch 34 mu var00.228215
compute loss for weight  0.790982  0.790972 result 0.554005
 training batch 35 mu var00.228215
compute loss for weight  0.790962  0.790972 result 0.553982
 training batch 36 mu var00.228215
compute loss for weight  0.790977  0.790972 result 0.554
 training batch 37 mu var00.228215
compute loss for weight  0.790967  0.790972 result 0.553988
   --dy = 1.16143 dy_ref = 1.16143
 training batch 38 mu var00.228215
compute loss for weight  0.510696  0.510686 result 0.553997
 training batch 39 mu var00.228215
compute loss for weight  0.510676  0.510686 result 0.55399
 training batch 40 mu var00.228215
compute loss for weight  0.510691  0.510686 result 0.553996
 training batch 41 mu var00.228215
compute loss for weight  0.510681  0.510686 result 0.553992
   --dy = 0.370739 dy_ref = 0.370739
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m1.31834e-10[NON-XML-CHAR-0x1B][39m