Execution Time0.12s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: master-x86_64-centos7-gcc48-dbg (olhswep22.cern.ch) on 2019-11-14 10:27:26

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 var00.633237
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      1.324       1.523 
   1 |    -0.8756      0.1947 
   2 |      2.501       1.262 
   3 |     0.1194     -0.1945 
   4 |      1.371       1.305 
   5 |      2.562       2.992 
   6 |     -1.287     -0.6868 
   7 |    -0.4722     -0.3648 
   8 |      1.328       2.121 
   9 |    -0.2372     -0.2669 

output BN 
output DL feature 0 mean 0.633237	output DL std 1.37113
output DL feature 1 mean 0.788561	output DL std 1.22943
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      0.531      0.6299 
   1 |      -1.16     -0.5091 
   2 |      1.436      0.4055 
   3 |     -0.395     -0.8429 
   4 |     0.5672      0.4431 
   5 |      1.483       1.889 
   6 |     -1.476      -1.265 
   7 |    -0.8498     -0.9888 
   8 |     0.5341       1.143 
   9 |    -0.6692     -0.9049 

output BN feature 0 mean 4.44089e-17	output BN std 1.05406
output BN feature 1 mean 1.44329e-16	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 |     0.1368    -0.00385     0.03544      0.1074 
   1 |     -3.189       1.298      -3.203      0.1557 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |     0.1922     -0.3554       1.225     -0.6591 
   1 |     0.9359      -0.128      0.6709     -0.2723 

 training batch 2 mu var00.63324
compute loss for weight  0.192214  0.192204 result 2.71097
 training batch 3 mu var00.633237
compute loss for weight  0.192194  0.192204 result 2.71097
 training batch 4 mu var00.633238
compute loss for weight  0.192209  0.192204 result 2.71097
 training batch 5 mu var00.633237
compute loss for weight  0.192199  0.192204 result 2.71097
   --dy = 0.136812 dy_ref = 0.136812
 training batch 6 mu var00.633237
compute loss for weight  -0.355373  -0.355383 result 2.71097
 training batch 7 mu var00.633237
compute loss for weight  -0.355393  -0.355383 result 2.71097
 training batch 8 mu var00.633237
compute loss for weight  -0.355378  -0.355383 result 2.71097
 training batch 9 mu var00.633237
compute loss for weight  -0.355388  -0.355383 result 2.71097
   --dy = -0.00385013 dy_ref = -0.00385013
 training batch 10 mu var00.633237
compute loss for weight  1.22488  1.22487 result 2.71097
 training batch 11 mu var00.633237
compute loss for weight  1.22486  1.22487 result 2.71097
 training batch 12 mu var00.633237
compute loss for weight  1.22487  1.22487 result 2.71097
 training batch 13 mu var00.633237
compute loss for weight  1.22486  1.22487 result 2.71097
   --dy = 0.0354432 dy_ref = 0.0354432
 training batch 14 mu var00.633237
compute loss for weight  -0.659087  -0.659097 result 2.71097
 training batch 15 mu var00.633237
compute loss for weight  -0.659107  -0.659097 result 2.71097
 training batch 16 mu var00.633237
compute loss for weight  -0.659092  -0.659097 result 2.71097
 training batch 17 mu var00.633237
compute loss for weight  -0.659102  -0.659097 result 2.71097
   --dy = 0.107411 dy_ref = 0.107411
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |       4.79      0.6317 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var00.633237
compute loss for weight  1.00001  1 result 2.71102
 training batch 19 mu var00.633237
compute loss for weight  0.99999  1 result 2.71092
 training batch 20 mu var00.633237
compute loss for weight  1.00001  1 result 2.71099
 training batch 21 mu var00.633237
compute loss for weight  0.999995  1 result 2.71095
   --dy = 4.7902 dy_ref = 4.7902
 training batch 22 mu var00.633237
compute loss for weight  1.00001  1 result 2.71098
 training batch 23 mu var00.633237
compute loss for weight  0.99999  1 result 2.71096
 training batch 24 mu var00.633237
compute loss for weight  1.00001  1 result 2.71097
 training batch 25 mu var00.633237
compute loss for weight  0.999995  1 result 2.71097
   --dy = 0.631745 dy_ref = 0.631745
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |  4.441e-16   6.245e-17 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var00.633237
compute loss for weight  1e-05  0 result 2.71097
 training batch 27 mu var00.633237
compute loss for weight  -1e-05  0 result 2.71097
 training batch 28 mu var00.633237
compute loss for weight  5e-06  0 result 2.71097
 training batch 29 mu var00.633237
compute loss for weight  -5e-06  0 result 2.71097
   --dy = 0 dy_ref = 4.44089e-16
 training batch 30 mu var00.633237
compute loss for weight  1e-05  0 result 2.71097
 training batch 31 mu var00.633237
compute loss for weight  -1e-05  0 result 2.71097
 training batch 32 mu var00.633237
compute loss for weight  5e-06  0 result 2.71097
 training batch 33 mu var00.633237
compute loss for weight  -5e-06  0 result 2.71097
   --dy = 5.18104e-11 dy_ref = 6.245e-17
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      3.287       2.992 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      1.457      0.2111 

 training batch 34 mu var00.633237
compute loss for weight  1.45738  1.45737 result 2.711
 training batch 35 mu var00.633237
compute loss for weight  1.45736  1.45737 result 2.71094
 training batch 36 mu var00.633237
compute loss for weight  1.45738  1.45737 result 2.71099
 training batch 37 mu var00.633237
compute loss for weight  1.45737  1.45737 result 2.71095
   --dy = 3.28688 dy_ref = 3.28688
 training batch 38 mu var00.633237
compute loss for weight  0.211156  0.211146 result 2.711
 training batch 39 mu var00.633237
compute loss for weight  0.211136  0.211146 result 2.71094
 training batch 40 mu var00.633237
compute loss for weight  0.211151  0.211146 result 2.71099
 training batch 41 mu var00.633237
compute loss for weight  0.211141  0.211146 result 2.71096
   --dy = 2.99198 dy_ref = 2.99198
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][33m2.70378e-09[NON-XML-CHAR-0x1B][39m