Execution Time0.52s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: master-x86_64-mac1015-clang110 (macphsft19.dyndns.cern.ch) on 2019-11-14 01:12:44

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 var00.51006
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      1.018      -1.103 
   1 |   -0.08842     -0.5178 
   2 |     0.5877      -1.043 
   3 |    -0.2435      0.3521 
   4 |     0.8638     -0.8527 
   5 |       1.93      -2.174 
   6 |     0.6831       1.909 
   7 |     -1.245      -1.328 
   8 |      1.734      -1.368 
   9 |     -0.138      0.2976 

output BN 
output DL feature 0 mean 0.51006	output DL std 0.963761
output DL feature 1 mean -0.582837	output DL std 1.16049
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.5558      -0.472 
   1 |    -0.6545     0.05911 
   2 |    0.08488      -0.418 
   3 |    -0.8242      0.8492 
   4 |     0.3868     -0.2451 
   5 |      1.553      -1.446 
   6 |     0.1892       2.263 
   7 |      -1.92     -0.6767 
   8 |      1.338     -0.7134 
   9 |    -0.7087      0.7997 

output BN feature 0 mean 1.11022e-16	output BN std 1.05403
output BN feature 1 mean -6.66134e-17	output BN std 1.05405
Testing weight gradients   for    layer 0
weight gradient for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |   0.006759    0.006892    0.007761   -0.009831 
   1 |      0.326       0.117      0.3587     -0.2418 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |     0.4233     -0.5074      0.4915      0.3125 
   1 |     -1.043     -0.6327     -0.4013      0.7476 

 training batch 2 mu var00.510063
compute loss for weight  0.423298  0.423288 result 0.448177
 training batch 3 mu var00.51006
compute loss for weight  0.423278  0.423288 result 0.448177
 training batch 4 mu var00.510061
compute loss for weight  0.423293  0.423288 result 0.448177
 training batch 5 mu var00.51006
compute loss for weight  0.423283  0.423288 result 0.448177
   --dy = 0.00675933 dy_ref = 0.00675933
 training batch 6 mu var00.51006
compute loss for weight  -0.507351  -0.507361 result 0.448177
 training batch 7 mu var00.51006
compute loss for weight  -0.507371  -0.507361 result 0.448177
 training batch 8 mu var00.51006
compute loss for weight  -0.507356  -0.507361 result 0.448177
 training batch 9 mu var00.51006
compute loss for weight  -0.507366  -0.507361 result 0.448177
   --dy = 0.00689249 dy_ref = 0.00689249
 training batch 10 mu var00.51006
compute loss for weight  0.491478  0.491468 result 0.448177
 training batch 11 mu var00.51006
compute loss for weight  0.491458  0.491468 result 0.448177
 training batch 12 mu var00.51006
compute loss for weight  0.491473  0.491468 result 0.448177
 training batch 13 mu var00.51006
compute loss for weight  0.491463  0.491468 result 0.448177
   --dy = 0.0077606 dy_ref = 0.0077606
 training batch 14 mu var00.51006
compute loss for weight  0.31249  0.31248 result 0.448177
 training batch 15 mu var00.51006
compute loss for weight  0.31247  0.31248 result 0.448177
 training batch 16 mu var00.51006
compute loss for weight  0.312485  0.31248 result 0.448177
 training batch 17 mu var00.51006
compute loss for weight  0.312475  0.31248 result 0.448177
   --dy = -0.00983057 dy_ref = -0.00983057
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.8895    0.006849 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var00.51006
compute loss for weight  1.00001  1 result 0.448186
 training batch 19 mu var00.51006
compute loss for weight  0.99999  1 result 0.448168
 training batch 20 mu var00.51006
compute loss for weight  1.00001  1 result 0.448182
 training batch 21 mu var00.51006
compute loss for weight  0.999995  1 result 0.448173
   --dy = 0.889505 dy_ref = 0.889505
 training batch 22 mu var00.51006
compute loss for weight  1.00001  1 result 0.448177
 training batch 23 mu var00.51006
compute loss for weight  0.99999  1 result 0.448177
 training batch 24 mu var00.51006
compute loss for weight  1.00001  1 result 0.448177
 training batch 25 mu var00.51006
compute loss for weight  0.999995  1 result 0.448177
   --dy = 0.00684913 dy_ref = 0.00684913
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |   1.11e-16  -3.036e-18 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var00.51006
compute loss for weight  1e-05  0 result 0.448177
 training batch 27 mu var00.51006
compute loss for weight  -1e-05  0 result 0.448177
 training batch 28 mu var00.51006
compute loss for weight  5e-06  0 result 0.448177
 training batch 29 mu var00.51006
compute loss for weight  -5e-06  0 result 0.448177
   --dy = 0 dy_ref = 1.11022e-16
 training batch 30 mu var00.51006
compute loss for weight  1e-05  0 result 0.448177
 training batch 31 mu var00.51006
compute loss for weight  -1e-05  0 result 0.448177
 training batch 32 mu var00.51006
compute loss for weight  5e-06  0 result 0.448177
 training batch 33 mu var00.51006
compute loss for weight  -5e-06  0 result 0.448177
   --dy = 1.4803e-11 dy_ref = -3.03577e-18
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      1.339     -0.4516 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.6645    -0.01517 

 training batch 34 mu var00.51006
compute loss for weight  0.664548  0.664538 result 0.448191
 training batch 35 mu var00.51006
compute loss for weight  0.664528  0.664538 result 0.448164
 training batch 36 mu var00.51006
compute loss for weight  0.664543  0.664538 result 0.448184
 training batch 37 mu var00.51006
compute loss for weight  0.664533  0.664538 result 0.44817
   --dy = 1.33853 dy_ref = 1.33853
 training batch 38 mu var00.51006
compute loss for weight  -0.0151564  -0.0151664 result 0.448173
 training batch 39 mu var00.51006
compute loss for weight  -0.0151764  -0.0151664 result 0.448182
 training batch 40 mu var00.51006
compute loss for weight  -0.0151614  -0.0151664 result 0.448175
 training batch 41 mu var00.51006
compute loss for weight  -0.0151714  -0.0151664 result 0.448179
   --dy = -0.451599 dy_ref = -0.451599
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][33m1.68339e-09[NON-XML-CHAR-0x1B][39m