Execution Time0.55s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: master-x86_64-mac1014-clang100-dbg (macphsft17.dyndns.cern.ch) on 2019-11-15 09:42:52

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 var00.0899591
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.5137     -0.2388 
   1 |     -1.466     0.01408 
   2 |      1.269      -0.268 
   3 |     -1.938     -0.6649 
   4 |     0.1961     -0.3801 
   5 |     0.3305     -0.6405 
   6 |     0.5464     -0.2136 
   7 |     0.0666      0.9559 
   8 |      1.747   0.0001896 
   9 |     -0.366    -0.08524 

output BN 
output DL feature 0 mean 0.0899591	output DL std 1.12231
output DL feature 1 mean -0.15208	output DL std 0.454744
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.3979      -0.201 
   1 |     -1.461      0.3851 
   2 |      1.108     -0.2686 
   3 |     -1.905      -1.188 
   4 |    0.09968     -0.5283 
   5 |     0.2259      -1.132 
   6 |     0.4287     -0.1425 
   7 |   -0.02194       2.568 
   8 |      1.556      0.3529 
   9 |    -0.4282      0.1549 

output BN feature 0 mean 5.55112e-18	output BN std 1.05405
output BN feature 1 mean 6.66134e-17	output BN std 1.05381
Testing weight gradients   for    layer 0
weight gradient for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |      0.299     -0.8721       0.183     -0.1095 
   1 |      6.756      -24.99     -0.0123       -5.68 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |    -0.4008     0.06754       1.222      0.4062 
   1 |   0.004392      0.4816    -0.04282     0.01075 

 training batch 2 mu var00.089962
compute loss for weight  -0.400819  -0.400829 result 2.48695
 training batch 3 mu var00.0899591
compute loss for weight  -0.400839  -0.400829 result 2.48694
 training batch 4 mu var00.0899599
compute loss for weight  -0.400824  -0.400829 result 2.48695
 training batch 5 mu var00.0899591
compute loss for weight  -0.400834  -0.400829 result 2.48694
   --dy = 0.298999 dy_ref = 0.298999
 training batch 6 mu var00.0899587
compute loss for weight  0.0675527  0.0675427 result 2.48694
 training batch 7 mu var00.0899591
compute loss for weight  0.0675327  0.0675427 result 2.48696
 training batch 8 mu var00.089959
compute loss for weight  0.0675477  0.0675427 result 2.48694
 training batch 9 mu var00.0899591
compute loss for weight  0.0675377  0.0675427 result 2.48695
   --dy = -0.872081 dy_ref = -0.872081
 training batch 10 mu var00.0899594
compute loss for weight  1.22237  1.22236 result 2.48695
 training batch 11 mu var00.0899591
compute loss for weight  1.22235  1.22236 result 2.48694
 training batch 12 mu var00.0899593
compute loss for weight  1.22236  1.22236 result 2.48695
 training batch 13 mu var00.0899591
compute loss for weight  1.22235  1.22236 result 2.48695
   --dy = 0.18296 dy_ref = 0.18296
 training batch 14 mu var00.0899591
compute loss for weight  0.406207  0.406197 result 2.48695
 training batch 15 mu var00.0899591
compute loss for weight  0.406187  0.406197 result 2.48695
 training batch 16 mu var00.0899591
compute loss for weight  0.406202  0.406197 result 2.48695
 training batch 17 mu var00.0899591
compute loss for weight  0.406192  0.406197 result 2.48695
   --dy = -0.109457 dy_ref = -0.109457
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      4.893     0.08111 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var00.0899591
compute loss for weight  1.00001  1 result 2.487
 training batch 19 mu var00.0899591
compute loss for weight  0.99999  1 result 2.4869
 training batch 20 mu var00.0899591
compute loss for weight  1.00001  1 result 2.48697
 training batch 21 mu var00.0899591
compute loss for weight  0.999995  1 result 2.48692
   --dy = 4.89278 dy_ref = 4.89278
 training batch 22 mu var00.0899591
compute loss for weight  1.00001  1 result 2.48695
 training batch 23 mu var00.0899591
compute loss for weight  0.99999  1 result 2.48695
 training batch 24 mu var00.0899591
compute loss for weight  1.00001  1 result 2.48695
 training batch 25 mu var00.0899591
compute loss for weight  0.999995  1 result 2.48695
   --dy = 0.0811115 dy_ref = 0.0811115
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 | -1.388e-16   6.939e-18 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var00.0899591
compute loss for weight  1e-05  0 result 2.48695
 training batch 27 mu var00.0899591
compute loss for weight  -1e-05  0 result 2.48695
 training batch 28 mu var00.0899591
compute loss for weight  5e-06  0 result 2.48695
 training batch 29 mu var00.0899591
compute loss for weight  -5e-06  0 result 2.48695
   --dy = -7.40149e-12 dy_ref = -1.38778e-16
 training batch 30 mu var00.0899591
compute loss for weight  1e-05  0 result 2.48695
 training batch 31 mu var00.0899591
compute loss for weight  -1e-05  0 result 2.48695
 training batch 32 mu var00.0899591
compute loss for weight  5e-06  0 result 2.48695
 training batch 33 mu var00.0899591
compute loss for weight  -5e-06  0 result 2.48695
   --dy = 5.92119e-11 dy_ref = 6.93889e-18
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      3.082     -0.2393 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      1.588      -0.339 

 training batch 34 mu var00.0899591
compute loss for weight  1.58775  1.58774 result 2.48698
 training batch 35 mu var00.0899591
compute loss for weight  1.58773  1.58774 result 2.48692
 training batch 36 mu var00.0899591
compute loss for weight  1.58775  1.58774 result 2.48696
 training batch 37 mu var00.0899591
compute loss for weight  1.58774  1.58774 result 2.48693
   --dy = 3.0816 dy_ref = 3.0816
 training batch 38 mu var00.0899591
compute loss for weight  -0.339003  -0.339013 result 2.48694
 training batch 39 mu var00.0899591
compute loss for weight  -0.339023  -0.339013 result 2.48695
 training batch 40 mu var00.0899591
compute loss for weight  -0.339008  -0.339013 result 2.48695
 training batch 41 mu var00.0899591
compute loss for weight  -0.339018  -0.339013 result 2.48695
   --dy = -0.239258 dy_ref = -0.239258
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m6.62602e-10[NON-XML-CHAR-0x1B][39m