Execution Time0.09s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: PR-4375-i686-ubuntu18-gcc7-opt (sft-ubuntu-1804-i386-2) on 2019-11-13 00:45:22
Repository revision: d782ec59ffdaa8482d351803546ea29ea22e1520

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 var00.322605
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.8451      0.2358 
   1 |      -1.43      0.3912 
   2 |      2.355      -1.176 
   3 |    -0.5597      -1.939 
   4 |     0.8483      -0.343 
   5 |      1.367    -0.01053 
   6 |    -0.8439       1.365 
   7 |    -0.1586      0.4802 
   8 |      1.068       1.871 
   9 |    -0.2652     -0.2605 

output BN 
output DL feature 0 mean 0.322605	output DL std 1.16027
output DL feature 1 mean 0.0613923	output DL std 1.11017
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.4747      0.1656 
   1 |     -1.592      0.3131 
   2 |      1.847      -1.175 
   3 |    -0.8016      -1.899 
   4 |     0.4776     -0.3839 
   5 |      0.949    -0.06829 
   6 |      -1.06       1.238 
   7 |    -0.4372      0.3976 
   8 |     0.6769       1.718 
   9 |     -0.534     -0.3056 

output BN feature 0 mean 4.44089e-17	output BN std 1.05405
output BN feature 1 mean 1.66533e-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 |   -0.02511     -0.0317     -0.0308    -0.06763 
   1 |   -0.07065      -0.057     -0.1147     -0.1274 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |    -0.3113     -0.2057       1.349     -0.4026 
   1 |     0.5259      0.4964    -0.05949      0.9073 

 training batch 2 mu var00.322608
compute loss for weight  -0.311318  -0.311328 result 0.119582
 training batch 3 mu var00.322605
compute loss for weight  -0.311338  -0.311328 result 0.119583
 training batch 4 mu var00.322605
compute loss for weight  -0.311323  -0.311328 result 0.119583
 training batch 5 mu var00.322605
compute loss for weight  -0.311333  -0.311328 result 0.119583
   --dy = -0.0251093 dy_ref = -0.0251093
 training batch 6 mu var00.322604
compute loss for weight  -0.205661  -0.205671 result 0.119582
 training batch 7 mu var00.322605
compute loss for weight  -0.205681  -0.205671 result 0.119583
 training batch 8 mu var00.322605
compute loss for weight  -0.205666  -0.205671 result 0.119583
 training batch 9 mu var00.322605
compute loss for weight  -0.205676  -0.205671 result 0.119583
   --dy = -0.0316994 dy_ref = -0.0316994
 training batch 10 mu var00.322605
compute loss for weight  1.34896  1.34895 result 0.119582
 training batch 11 mu var00.322605
compute loss for weight  1.34894  1.34895 result 0.119583
 training batch 12 mu var00.322605
compute loss for weight  1.34895  1.34895 result 0.119583
 training batch 13 mu var00.322605
compute loss for weight  1.34894  1.34895 result 0.119583
   --dy = -0.0308033 dy_ref = -0.0308033
 training batch 14 mu var00.322605
compute loss for weight  -0.402607  -0.402617 result 0.119582
 training batch 15 mu var00.322605
compute loss for weight  -0.402627  -0.402617 result 0.119583
 training batch 16 mu var00.322605
compute loss for weight  -0.402612  -0.402617 result 0.119582
 training batch 17 mu var00.322605
compute loss for weight  -0.402622  -0.402617 result 0.119583
   --dy = -0.0676299 dy_ref = -0.0676299
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     0.1667     0.07242 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var00.322605
compute loss for weight  1.00001  1 result 0.119584
 training batch 19 mu var00.322605
compute loss for weight  0.99999  1 result 0.119581
 training batch 20 mu var00.322605
compute loss for weight  1.00001  1 result 0.119584
 training batch 21 mu var00.322605
compute loss for weight  0.999995  1 result 0.119582
   --dy = 0.166748 dy_ref = 0.166748
 training batch 22 mu var00.322605
compute loss for weight  1.00001  1 result 0.119583
 training batch 23 mu var00.322605
compute loss for weight  0.99999  1 result 0.119582
 training batch 24 mu var00.322605
compute loss for weight  1.00001  1 result 0.119583
 training batch 25 mu var00.322605
compute loss for weight  0.999995  1 result 0.119582
   --dy = 0.0724174 dy_ref = 0.0724174
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |  1.388e-17  -6.939e-18 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var00.322605
compute loss for weight  1e-05  0 result 0.119583
 training batch 27 mu var00.322605
compute loss for weight  -1e-05  0 result 0.119583
 training batch 28 mu var00.322605
compute loss for weight  5e-06  0 result 0.119583
 training batch 29 mu var00.322605
compute loss for weight  -5e-06  0 result 0.119583
   --dy = 4.62593e-13 dy_ref = 1.38778e-17
 training batch 30 mu var00.322605
compute loss for weight  1e-05  0 result 0.119583
 training batch 31 mu var00.322605
compute loss for weight  -1e-05  0 result 0.119583
 training batch 32 mu var00.322605
compute loss for weight  5e-06  0 result 0.119583
 training batch 33 mu var00.322605
compute loss for weight  -5e-06  0 result 0.119583
   --dy = -4.62593e-13 dy_ref = -6.93889e-18
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.6033      0.4235 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.2764       0.171 

 training batch 34 mu var00.322605
compute loss for weight  -0.276406  -0.276416 result 0.119577
 training batch 35 mu var00.322605
compute loss for weight  -0.276426  -0.276416 result 0.119589
 training batch 36 mu var00.322605
compute loss for weight  -0.276411  -0.276416 result 0.11958
 training batch 37 mu var00.322605
compute loss for weight  -0.276421  -0.276416 result 0.119586
   --dy = -0.60325 dy_ref = -0.60325
 training batch 38 mu var00.322605
compute loss for weight  0.170998  0.170988 result 0.119587
 training batch 39 mu var00.322605
compute loss for weight  0.170978  0.170988 result 0.119578
 training batch 40 mu var00.322605
compute loss for weight  0.170993  0.170988 result 0.119585
 training batch 41 mu var00.322605
compute loss for weight  0.170983  0.170988 result 0.119581
   --dy = 0.423524 dy_ref = 0.423524
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m1.27991e-10[NON-XML-CHAR-0x1B][39m