Execution Time1.10s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: PR-4624-x86_64-centos7-gcc48-opt (olhswep22.cern.ch) on 2019-11-14 18:16:07

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 var00.697614
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      1.317     -0.8572 
   1 |     0.3442     -0.5459 
   2 |      1.109     -0.7983 
   3 |   -0.04616     -0.6193 
   4 |      1.138     -0.8612 
   5 |      2.646      -1.938 
   6 |     -1.024       1.328 
   7 |    0.03884     -0.1988 
   8 |      1.685     -0.6057 
   9 |    -0.2322     0.07926 

output BN 
output DL feature 0 mean 0.697614	output DL std 1.078
output DL feature 1 mean -0.501694	output DL std 0.830662
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.6059     -0.4511 
   1 |    -0.3456    -0.05607 
   2 |     0.4023     -0.3764 
   3 |    -0.7272     -0.1493 
   4 |     0.4305     -0.4561 
   5 |      1.905      -1.822 
   6 |     -1.683       2.322 
   7 |    -0.6441      0.3844 
   8 |     0.9654      -0.132 
   9 |    -0.9091      0.7372 

output BN feature 0 mean 3.33067e-17	output BN std 1.05404
output BN feature 1 mean 5.55112e-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 |    -0.2336     -0.2547     -0.8024      -1.481 
   1 |       3.29      -2.151       1.496      -5.896 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |     0.9308     0.04132      0.5023     -0.4263 
   1 |    -0.7573    -0.05686     -0.1361      0.6359 

 training batch 2 mu var00.697616
compute loss for weight  0.930769  0.930759 result 10.9038
 training batch 3 mu var00.697614
compute loss for weight  0.930749  0.930759 result 10.9038
 training batch 4 mu var00.697614
compute loss for weight  0.930764  0.930759 result 10.9038
 training batch 5 mu var00.697614
compute loss for weight  0.930754  0.930759 result 10.9038
   --dy = -0.233551 dy_ref = -0.233551
 training batch 6 mu var00.697613
compute loss for weight  0.0413258  0.0413158 result 10.9038
 training batch 7 mu var00.697614
compute loss for weight  0.0413058  0.0413158 result 10.9038
 training batch 8 mu var00.697613
compute loss for weight  0.0413208  0.0413158 result 10.9038
 training batch 9 mu var00.697614
compute loss for weight  0.0413108  0.0413158 result 10.9038
   --dy = -0.254746 dy_ref = -0.254746
 training batch 10 mu var00.697614
compute loss for weight  0.502288  0.502278 result 10.9038
 training batch 11 mu var00.697614
compute loss for weight  0.502268  0.502278 result 10.9038
 training batch 12 mu var00.697614
compute loss for weight  0.502283  0.502278 result 10.9038
 training batch 13 mu var00.697614
compute loss for weight  0.502273  0.502278 result 10.9038
   --dy = -0.802436 dy_ref = -0.802436
 training batch 14 mu var00.697613
compute loss for weight  -0.426248  -0.426258 result 10.9038
 training batch 15 mu var00.697614
compute loss for weight  -0.426268  -0.426258 result 10.9038
 training batch 16 mu var00.697614
compute loss for weight  -0.426253  -0.426258 result 10.9038
 training batch 17 mu var00.697614
compute loss for weight  -0.426263  -0.426258 result 10.9038
   --dy = -1.4813 dy_ref = -1.4813
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      4.874       16.93 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var00.697614
compute loss for weight  1.00001  1 result 10.9039
 training batch 19 mu var00.697614
compute loss for weight  0.99999  1 result 10.9038
 training batch 20 mu var00.697614
compute loss for weight  1.00001  1 result 10.9038
 training batch 21 mu var00.697614
compute loss for weight  0.999995  1 result 10.9038
   --dy = 4.87436 dy_ref = 4.87436
 training batch 22 mu var00.697614
compute loss for weight  1.00001  1 result 10.904
 training batch 23 mu var00.697614
compute loss for weight  0.99999  1 result 10.9037
 training batch 24 mu var00.697614
compute loss for weight  1.00001  1 result 10.9039
 training batch 25 mu var00.697614
compute loss for weight  0.999995  1 result 10.9037
   --dy = 16.9333 dy_ref = 16.9333
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 | -3.331e-16   4.441e-16 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var00.697614
compute loss for weight  1e-05  0 result 10.9038
 training batch 27 mu var00.697614
compute loss for weight  -1e-05  0 result 10.9038
 training batch 28 mu var00.697614
compute loss for weight  5e-06  0 result 10.9038
 training batch 29 mu var00.697614
compute loss for weight  -5e-06  0 result 10.9038
   --dy = -2.66454e-10 dy_ref = -3.33067e-16
 training batch 30 mu var00.697614
compute loss for weight  1e-05  0 result 10.9038
 training batch 31 mu var00.697614
compute loss for weight  -1e-05  0 result 10.9038
 training batch 32 mu var00.697614
compute loss for weight  5e-06  0 result 10.9038
 training batch 33 mu var00.697614
compute loss for weight  -5e-06  0 result 10.9038
   --dy = -2.66454e-10 dy_ref = 4.44089e-16
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      6.178      -6.565 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.7889      -2.579 

 training batch 34 mu var00.697614
compute loss for weight  0.788953  0.788943 result 10.9039
 training batch 35 mu var00.697614
compute loss for weight  0.788933  0.788943 result 10.9038
 training batch 36 mu var00.697614
compute loss for weight  0.788948  0.788943 result 10.9039
 training batch 37 mu var00.697614
compute loss for weight  0.788938  0.788943 result 10.9038
   --dy = 6.17834 dy_ref = 6.17834
 training batch 38 mu var00.697614
compute loss for weight  -2.57931  -2.57932 result 10.9038
 training batch 39 mu var00.697614
compute loss for weight  -2.57933  -2.57932 result 10.9039
 training batch 40 mu var00.697614
compute loss for weight  -2.57932  -2.57932 result 10.9038
 training batch 41 mu var00.697614
compute loss for weight  -2.57933  -2.57932 result 10.9039
   --dy = -6.56501 dy_ref = -6.56501
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][33m1.32892e-09[NON-XML-CHAR-0x1B][39m