Execution Time1.09s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: PR-4623-x86_64-fedora27-gcc7-opt (sft-fedora-27-2.cern.ch) on 2019-11-14 10:11:28
Repository revision: 40f17f5d2096bfd8b3844c3195a20ad01a6fb2eb

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 var0-0.103658
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.2697     -0.5573 
   1 |    -0.0234      0.2493 
   2 |     0.7684      -1.257 
   3 |     0.8864      -1.357 
   4 |   0.002829     -0.8846 
   5 |    -0.2793      -1.434 
   6 |     -1.933      0.6431 
   7 |      1.124       1.176 
   8 |     -1.385      0.2086 
   9 |     0.0723     -0.1038 

output BN 
output DL feature 0 mean -0.103658	output DL std 0.962715
output DL feature 1 mean -0.331651	output DL std 0.905733
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.1818     -0.2626 
   1 |    0.08787       0.676 
   2 |     0.9547      -1.077 
   3 |      1.084      -1.193 
   4 |     0.1166     -0.6434 
   5 |    -0.1923      -1.283 
   6 |     -2.003       1.134 
   7 |      1.344       1.755 
   8 |     -1.403      0.6287 
   9 |     0.1926      0.2652 

output BN feature 0 mean -5.55112e-18	output BN std 1.05403
output BN feature 1 mean 3.88578e-17	output BN std 1.05402
Testing weight gradients   for    layer 0
weight gradient for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |    0.01101    -0.02668    0.009643     -0.0112 
   1 |    0.02156    -0.04112     0.01808     0.00407 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |    -0.1259       0.343    -0.03059     -0.9672 
   1 |     0.0268      0.7002     -0.3383      0.5068 

 training batch 2 mu var0-0.103655
compute loss for weight  -0.125882  -0.125892 result 0.0425358
 training batch 3 mu var0-0.103658
compute loss for weight  -0.125902  -0.125892 result 0.0425355
 training batch 4 mu var0-0.103657
compute loss for weight  -0.125887  -0.125892 result 0.0425357
 training batch 5 mu var0-0.103658
compute loss for weight  -0.125897  -0.125892 result 0.0425356
   --dy = 0.0110082 dy_ref = 0.0110082
 training batch 6 mu var0-0.103658
compute loss for weight  0.34299  0.34298 result 0.0425354
 training batch 7 mu var0-0.103658
compute loss for weight  0.34297  0.34298 result 0.0425359
 training batch 8 mu var0-0.103658
compute loss for weight  0.342985  0.34298 result 0.0425355
 training batch 9 mu var0-0.103658
compute loss for weight  0.342975  0.34298 result 0.0425358
   --dy = -0.0266773 dy_ref = -0.0266773
 training batch 10 mu var0-0.103658
compute loss for weight  -0.0305832  -0.0305932 result 0.0425358
 training batch 11 mu var0-0.103658
compute loss for weight  -0.0306032  -0.0305932 result 0.0425356
 training batch 12 mu var0-0.103658
compute loss for weight  -0.0305882  -0.0305932 result 0.0425357
 training batch 13 mu var0-0.103658
compute loss for weight  -0.0305982  -0.0305932 result 0.0425356
   --dy = 0.00964326 dy_ref = 0.00964326
 training batch 14 mu var0-0.103658
compute loss for weight  -0.967235  -0.967245 result 0.0425355
 training batch 15 mu var0-0.103658
compute loss for weight  -0.967255  -0.967245 result 0.0425358
 training batch 16 mu var0-0.103658
compute loss for weight  -0.96724  -0.967245 result 0.0425356
 training batch 17 mu var0-0.103658
compute loss for weight  -0.96725  -0.967245 result 0.0425357
   --dy = -0.0112012 dy_ref = -0.0112012
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    0.03037      0.0547 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var0-0.103658
compute loss for weight  1.00001  1 result 0.042536
 training batch 19 mu var0-0.103658
compute loss for weight  0.99999  1 result 0.0425354
 training batch 20 mu var0-0.103658
compute loss for weight  1.00001  1 result 0.0425358
 training batch 21 mu var0-0.103658
compute loss for weight  0.999995  1 result 0.0425355
   --dy = 0.0303666 dy_ref = 0.0303666
 training batch 22 mu var0-0.103658
compute loss for weight  1.00001  1 result 0.0425362
 training batch 23 mu var0-0.103658
compute loss for weight  0.99999  1 result 0.0425351
 training batch 24 mu var0-0.103658
compute loss for weight  1.00001  1 result 0.0425359
 training batch 25 mu var0-0.103658
compute loss for weight  0.999995  1 result 0.0425354
   --dy = 0.0547047 dy_ref = 0.0547047
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |  2.656e-18  -4.879e-18 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var0-0.103658
compute loss for weight  1e-05  0 result 0.0425357
 training batch 27 mu var0-0.103658
compute loss for weight  -1e-05  0 result 0.0425357
 training batch 28 mu var0-0.103658
compute loss for weight  5e-06  0 result 0.0425357
 training batch 29 mu var0-0.103658
compute loss for weight  -5e-06  0 result 0.0425357
   --dy = 1.15648e-12 dy_ref = 2.6563e-18
 training batch 30 mu var0-0.103658
compute loss for weight  1e-05  0 result 0.0425357
 training batch 31 mu var0-0.103658
compute loss for weight  -1e-05  0 result 0.0425357
 training batch 32 mu var0-0.103658
compute loss for weight  5e-06  0 result 0.0425357
 training batch 33 mu var0-0.103658
compute loss for weight  -5e-06  0 result 0.0425357
   --dy = -9.25186e-13 dy_ref = -4.87891e-18
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.2923       0.361 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.1039      0.1515 

 training batch 34 mu var0-0.103658
compute loss for weight  -0.103892  -0.103902 result 0.0425327
 training batch 35 mu var0-0.103658
compute loss for weight  -0.103912  -0.103902 result 0.0425386
 training batch 36 mu var0-0.103658
compute loss for weight  -0.103897  -0.103902 result 0.0425342
 training batch 37 mu var0-0.103658
compute loss for weight  -0.103907  -0.103902 result 0.0425371
   --dy = -0.292263 dy_ref = -0.292263
 training batch 38 mu var0-0.103658
compute loss for weight  0.151557  0.151547 result 0.0425393
 training batch 39 mu var0-0.103658
compute loss for weight  0.151537  0.151547 result 0.042532
 training batch 40 mu var0-0.103658
compute loss for weight  0.151552  0.151547 result 0.0425375
 training batch 41 mu var0-0.103658
compute loss for weight  0.151542  0.151547 result 0.0425339
   --dy = 0.360976 dy_ref = 0.360976
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m2.43607e-10[NON-XML-CHAR-0x1B][39m