Execution Time0.18s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: PR-4623-i686-ubuntu18-gcc7-opt (sft-ubuntu-1804-i386-2) on 2019-11-14 10:48:29
Repository revision: 14de58de35eff907054671888ccc2de0f7f27e77

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 var00.937593
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      1.584     -0.6154 
   1 |     0.7568     0.08567 
   2 |     0.9262      -1.662 
   3 |      1.664     -0.4138 
   4 |      1.731     -0.7003 
   5 |      3.656      -1.347 
   6 |    -0.1383       2.482 
   7 |     -2.216      -1.266 
   8 |      1.354    0.009568 
   9 |    0.05709      0.1503 

output BN 
output DL feature 0 mean 0.937593	output DL std 1.52921
output DL feature 1 mean -0.32769	output DL std 1.17085
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.4459      -0.259 
   1 |    -0.1246      0.3721 
   2 |  -0.007846      -1.201 
   3 |     0.5008    -0.07751 
   4 |     0.5469     -0.3354 
   5 |      1.874     -0.9179 
   6 |    -0.7416        2.53 
   7 |     -2.173     -0.8451 
   8 |     0.2871      0.3036 
   9 |    -0.6069      0.4303 

output BN feature 0 mean 1.44329e-16	output BN std 1.05407
output BN feature 1 mean 2.22045e-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 |    -0.1622     -0.8504     -0.8952       2.165 
   1 |      3.403      -2.616       3.229       -4.62 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |     0.9973      -1.046      0.1605     -0.2695 
   1 |    -0.3749     -0.4353     -0.5061       1.151 

 training batch 2 mu var00.937596
compute loss for weight  0.9973  0.99729 result 3.84294
 training batch 3 mu var00.937593
compute loss for weight  0.99728  0.99729 result 3.84295
 training batch 4 mu var00.937594
compute loss for weight  0.997295  0.99729 result 3.84295
 training batch 5 mu var00.937593
compute loss for weight  0.997285  0.99729 result 3.84295
   --dy = -0.162235 dy_ref = -0.162235
 training batch 6 mu var00.937593
compute loss for weight  -1.04558  -1.04559 result 3.84294
 training batch 7 mu var00.937593
compute loss for weight  -1.0456  -1.04559 result 3.84295
 training batch 8 mu var00.937593
compute loss for weight  -1.04558  -1.04559 result 3.84294
 training batch 9 mu var00.937593
compute loss for weight  -1.04559  -1.04559 result 3.84295
   --dy = -0.850367 dy_ref = -0.850367
 training batch 10 mu var00.937593
compute loss for weight  0.160558  0.160548 result 3.84294
 training batch 11 mu var00.937593
compute loss for weight  0.160538  0.160548 result 3.84296
 training batch 12 mu var00.937593
compute loss for weight  0.160553  0.160548 result 3.84294
 training batch 13 mu var00.937593
compute loss for weight  0.160543  0.160548 result 3.84295
   --dy = -0.895215 dy_ref = -0.895215
 training batch 14 mu var00.937593
compute loss for weight  -0.269493  -0.269503 result 3.84297
 training batch 15 mu var00.937593
compute loss for weight  -0.269513  -0.269503 result 3.84292
 training batch 16 mu var00.937593
compute loss for weight  -0.269498  -0.269503 result 3.84296
 training batch 17 mu var00.937593
compute loss for weight  -0.269508  -0.269503 result 3.84294
   --dy = 2.16455 dy_ref = 2.16455
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      5.427       2.259 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var00.937593
compute loss for weight  1.00001  1 result 3.843
 training batch 19 mu var00.937593
compute loss for weight  0.99999  1 result 3.84289
 training batch 20 mu var00.937593
compute loss for weight  1.00001  1 result 3.84297
 training batch 21 mu var00.937593
compute loss for weight  0.999995  1 result 3.84292
   --dy = 5.42687 dy_ref = 5.42687
 training batch 22 mu var00.937593
compute loss for weight  1.00001  1 result 3.84297
 training batch 23 mu var00.937593
compute loss for weight  0.99999  1 result 3.84292
 training batch 24 mu var00.937593
compute loss for weight  1.00001  1 result 3.84296
 training batch 25 mu var00.937593
compute loss for weight  0.999995  1 result 3.84294
   --dy = 2.25903 dy_ref = 2.25903
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |  1.138e-15   5.829e-16 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var00.937593
compute loss for weight  1e-05  0 result 3.84295
 training batch 27 mu var00.937593
compute loss for weight  -1e-05  0 result 3.84295
 training batch 28 mu var00.937593
compute loss for weight  5e-06  0 result 3.84295
 training batch 29 mu var00.937593
compute loss for weight  -5e-06  0 result 3.84295
   --dy = 5.92119e-11 dy_ref = 1.13798e-15
 training batch 30 mu var00.937593
compute loss for weight  1e-05  0 result 3.84295
 training batch 31 mu var00.937593
compute loss for weight  -1e-05  0 result 3.84295
 training batch 32 mu var00.937593
compute loss for weight  5e-06  0 result 3.84295
 training batch 33 mu var00.937593
compute loss for weight  -5e-06  0 result 3.84295
   --dy = -1.18424e-10 dy_ref = 5.82867e-16
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      -3.01       -1.75 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     -1.803      -1.291 

 training batch 34 mu var00.937593
compute loss for weight  -1.80298  -1.80299 result 3.84292
 training batch 35 mu var00.937593
compute loss for weight  -1.803  -1.80299 result 3.84298
 training batch 36 mu var00.937593
compute loss for weight  -1.80299  -1.80299 result 3.84293
 training batch 37 mu var00.937593
compute loss for weight  -1.803  -1.80299 result 3.84296
   --dy = -3.00992 dy_ref = -3.00992
 training batch 38 mu var00.937593
compute loss for weight  -1.29105  -1.29106 result 3.84293
 training batch 39 mu var00.937593
compute loss for weight  -1.29107  -1.29106 result 3.84296
 training batch 40 mu var00.937593
compute loss for weight  -1.29106  -1.29106 result 3.84294
 training batch 41 mu var00.937593
compute loss for weight  -1.29107  -1.29106 result 3.84296
   --dy = -1.74974 dy_ref = -1.74974
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m5.37289e-10[NON-XML-CHAR-0x1B][39m