Execution Time0.46s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: master-x86_64-mac1015-clang110 (macphsft19.dyndns.cern.ch) on 2019-11-12 01:29:41

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 var00.388481
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.6178      0.4061 
   1 |      1.005    -0.09575 
   2 |    -0.7461     -0.2563 
   3 |     0.1636      -1.828 
   4 |     0.4399    -0.08099 
   5 |      1.374      0.3062 
   6 |     0.6405      0.7083 
   7 |    -0.7055      0.5828 
   8 |      1.104         1.8 
   9 |  -0.008027     -0.3072 

output BN 
output DL feature 0 mean 0.388481	output DL std 0.720609
output DL feature 1 mean 0.123536	output DL std 0.925232
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.3354      0.3218 
   1 |     0.9021     -0.2498 
   2 |      -1.66     -0.4327 
   3 |     -0.329      -2.223 
   4 |    0.07513      -0.233 
   5 |      1.441      0.2081 
   6 |     0.3686      0.6662 
   7 |       -1.6      0.5232 
   8 |      1.046        1.91 
   9 |    -0.5799     -0.4907 

output BN feature 0 mean 4.44089e-17	output BN std 1.05398
output BN feature 1 mean 5.55112e-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 |     0.1726      0.7677      0.6951      0.7784 
   1 |    0.03346      -1.249     -0.9271     -0.9038 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |     0.8356     -0.1826     -0.3666      0.3226 
   1 |     0.3477      0.4531      0.3489      0.5667 

 training batch 2 mu var00.388484
compute loss for weight  0.835619  0.835609 result 1.35996
 training batch 3 mu var00.388481
compute loss for weight  0.835599  0.835609 result 1.35996
 training batch 4 mu var00.388482
compute loss for weight  0.835614  0.835609 result 1.35996
 training batch 5 mu var00.388481
compute loss for weight  0.835604  0.835609 result 1.35996
   --dy = 0.172604 dy_ref = 0.172604
 training batch 6 mu var00.38848
compute loss for weight  -0.182579  -0.182589 result 1.35997
 training batch 7 mu var00.388481
compute loss for weight  -0.182599  -0.182589 result 1.35995
 training batch 8 mu var00.388481
compute loss for weight  -0.182584  -0.182589 result 1.35996
 training batch 9 mu var00.388481
compute loss for weight  -0.182594  -0.182589 result 1.35996
   --dy = 0.767727 dy_ref = 0.767727
 training batch 10 mu var00.388481
compute loss for weight  -0.366553  -0.366563 result 1.35997
 training batch 11 mu var00.388481
compute loss for weight  -0.366573  -0.366563 result 1.35995
 training batch 12 mu var00.388481
compute loss for weight  -0.366558  -0.366563 result 1.35996
 training batch 13 mu var00.388481
compute loss for weight  -0.366568  -0.366563 result 1.35996
   --dy = 0.695055 dy_ref = 0.695055
 training batch 14 mu var00.388481
compute loss for weight  0.322618  0.322608 result 1.35997
 training batch 15 mu var00.388481
compute loss for weight  0.322598  0.322608 result 1.35995
 training batch 16 mu var00.388481
compute loss for weight  0.322613  0.322608 result 1.35996
 training batch 17 mu var00.388481
compute loss for weight  0.322603  0.322608 result 1.35996
   --dy = 0.778381 dy_ref = 0.778381
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      1.785      0.9352 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var00.388481
compute loss for weight  1.00001  1 result 1.35998
 training batch 19 mu var00.388481
compute loss for weight  0.99999  1 result 1.35994
 training batch 20 mu var00.388481
compute loss for weight  1.00001  1 result 1.35997
 training batch 21 mu var00.388481
compute loss for weight  0.999995  1 result 1.35995
   --dy = 1.78469 dy_ref = 1.78469
 training batch 22 mu var00.388481
compute loss for weight  1.00001  1 result 1.35997
 training batch 23 mu var00.388481
compute loss for weight  0.99999  1 result 1.35995
 training batch 24 mu var00.388481
compute loss for weight  1.00001  1 result 1.35996
 training batch 25 mu var00.388481
compute loss for weight  0.999995  1 result 1.35995
   --dy = 0.935226 dy_ref = 0.935226
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |  1.388e-16   9.714e-17 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var00.388481
compute loss for weight  1e-05  0 result 1.35996
 training batch 27 mu var00.388481
compute loss for weight  -1e-05  0 result 1.35996
 training batch 28 mu var00.388481
compute loss for weight  5e-06  0 result 1.35996
 training batch 29 mu var00.388481
compute loss for weight  -5e-06  0 result 1.35996
   --dy = -3.70074e-12 dy_ref = 1.38778e-16
 training batch 30 mu var00.388481
compute loss for weight  1e-05  0 result 1.35996
 training batch 31 mu var00.388481
compute loss for weight  -1e-05  0 result 1.35996
 training batch 32 mu var00.388481
compute loss for weight  5e-06  0 result 1.35996
 training batch 33 mu var00.388481
compute loss for weight  -5e-06  0 result 1.35996
   --dy = 3.70074e-12 dy_ref = 9.71445e-17
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     -2.082       -1.68 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.8572     -0.5567 

 training batch 34 mu var00.388481
compute loss for weight  -0.857184  -0.857194 result 1.35994
 training batch 35 mu var00.388481
compute loss for weight  -0.857204  -0.857194 result 1.35998
 training batch 36 mu var00.388481
compute loss for weight  -0.857189  -0.857194 result 1.35995
 training batch 37 mu var00.388481
compute loss for weight  -0.857199  -0.857194 result 1.35997
   --dy = -2.08202 dy_ref = -2.08202
 training batch 38 mu var00.388481
compute loss for weight  -0.556705  -0.556715 result 1.35994
 training batch 39 mu var00.388481
compute loss for weight  -0.556725  -0.556715 result 1.35998
 training batch 40 mu var00.388481
compute loss for weight  -0.55671  -0.556715 result 1.35995
 training batch 41 mu var00.388481
compute loss for weight  -0.55672  -0.556715 result 1.35997
   --dy = -1.6799 dy_ref = -1.6799
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m1.61188e-10[NON-XML-CHAR-0x1B][39m