Execution Time1.04s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: PR-420-x86_64-centos7-gcc48-opt (olhswep22.cern.ch) on 2019-11-14 09:40:23

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 var0-0.548324
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      -1.08      -1.211 
   1 |    0.06459      -1.824 
   2 |     -1.239     -0.2007 
   3 |    0.09758      -3.718 
   4 |    -0.9702       -1.75 
   5 |     -2.114      -3.585 
   6 |     0.8099      0.6815 
   7 |     0.1113        2.69 
   8 |     -1.366      0.3281 
   9 |      0.203     -0.4177 

output BN 
output DL feature 0 mean -0.548324	output DL std 0.924633
output DL feature 1 mean -0.90064	output DL std 1.95631
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.6066     -0.1672 
   1 |     0.6987     -0.4976 
   2 |    -0.7872      0.3772 
   3 |     0.7363      -1.518 
   4 |    -0.4809     -0.4575 
   5 |     -1.784      -1.447 
   6 |      1.548      0.8525 
   7 |      0.752       1.935 
   8 |    -0.9326      0.6621 
   9 |     0.8564      0.2602 

output BN feature 0 mean 0	output BN std 1.05402
output BN feature 1 mean -1.55431e-16	output BN std 1.05408
Testing weight gradients   for    layer 0
weight gradient for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |     0.3409      -1.794      -1.013     -0.9015 
   1 |     0.7692     -0.3478      0.8997     -0.2696 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |    -0.5599      0.0641     -0.6174      0.3545 
   1 |     -1.208       1.351      0.7743      0.7239 

 training batch 2 mu var0-0.548321
compute loss for weight  -0.559875  -0.559885 result 1.52423
 training batch 3 mu var0-0.548324
compute loss for weight  -0.559895  -0.559885 result 1.52422
 training batch 4 mu var0-0.548323
compute loss for weight  -0.55988  -0.559885 result 1.52423
 training batch 5 mu var0-0.548324
compute loss for weight  -0.55989  -0.559885 result 1.52423
   --dy = 0.340891 dy_ref = 0.340891
 training batch 6 mu var0-0.548324
compute loss for weight  0.0641052  0.0640952 result 1.52421
 training batch 7 mu var0-0.548324
compute loss for weight  0.0640852  0.0640952 result 1.52425
 training batch 8 mu var0-0.548324
compute loss for weight  0.0641002  0.0640952 result 1.52422
 training batch 9 mu var0-0.548324
compute loss for weight  0.0640902  0.0640952 result 1.52424
   --dy = -1.79353 dy_ref = -1.79353
 training batch 10 mu var0-0.548324
compute loss for weight  -0.617397  -0.617407 result 1.52422
 training batch 11 mu var0-0.548324
compute loss for weight  -0.617417  -0.617407 result 1.52424
 training batch 12 mu var0-0.548324
compute loss for weight  -0.617402  -0.617407 result 1.52422
 training batch 13 mu var0-0.548324
compute loss for weight  -0.617412  -0.617407 result 1.52423
   --dy = -1.01339 dy_ref = -1.01339
 training batch 14 mu var0-0.548324
compute loss for weight  0.354527  0.354517 result 1.52422
 training batch 15 mu var0-0.548324
compute loss for weight  0.354507  0.354517 result 1.52424
 training batch 16 mu var0-0.548324
compute loss for weight  0.354522  0.354517 result 1.52422
 training batch 17 mu var0-0.548324
compute loss for weight  0.354512  0.354517 result 1.52423
   --dy = -0.901537 dy_ref = -0.901537
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |        1.9       1.148 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var0-0.548324
compute loss for weight  1.00001  1 result 1.52425
 training batch 19 mu var0-0.548324
compute loss for weight  0.99999  1 result 1.52421
 training batch 20 mu var0-0.548324
compute loss for weight  1.00001  1 result 1.52424
 training batch 21 mu var0-0.548324
compute loss for weight  0.999995  1 result 1.52422
   --dy = 1.90003 dy_ref = 1.90003
 training batch 22 mu var0-0.548324
compute loss for weight  1.00001  1 result 1.52424
 training batch 23 mu var0-0.548324
compute loss for weight  0.99999  1 result 1.52422
 training batch 24 mu var0-0.548324
compute loss for weight  1.00001  1 result 1.52423
 training batch 25 mu var0-0.548324
compute loss for weight  0.999995  1 result 1.52422
   --dy = 1.14843 dy_ref = 1.14843
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |  1.665e-16  -1.665e-16 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var0-0.548324
compute loss for weight  1e-05  0 result 1.52423
 training batch 27 mu var0-0.548324
compute loss for weight  -1e-05  0 result 1.52423
 training batch 28 mu var0-0.548324
compute loss for weight  5e-06  0 result 1.52423
 training batch 29 mu var0-0.548324
compute loss for weight  -5e-06  0 result 1.52423
   --dy = 2.96059e-11 dy_ref = 1.66533e-16
 training batch 30 mu var0-0.548324
compute loss for weight  1e-05  0 result 1.52423
 training batch 31 mu var0-0.548324
compute loss for weight  -1e-05  0 result 1.52423
 training batch 32 mu var0-0.548324
compute loss for weight  5e-06  0 result 1.52423
 training batch 33 mu var0-0.548324
compute loss for weight  -5e-06  0 result 1.52423
   --dy = 2.59052e-11 dy_ref = -1.66533e-16
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     -1.632        1.16 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     -1.164        0.99 

 training batch 34 mu var0-0.548324
compute loss for weight  -1.16448  -1.16449 result 1.52421
 training batch 35 mu var0-0.548324
compute loss for weight  -1.1645  -1.16449 result 1.52424
 training batch 36 mu var0-0.548324
compute loss for weight  -1.16448  -1.16449 result 1.52422
 training batch 37 mu var0-0.548324
compute loss for weight  -1.16449  -1.16449 result 1.52424
   --dy = -1.63164 dy_ref = -1.63164
 training batch 38 mu var0-0.548324
compute loss for weight  0.99  0.98999 result 1.52424
 training batch 39 mu var0-0.548324
compute loss for weight  0.98998  0.98999 result 1.52422
 training batch 40 mu var0-0.548324
compute loss for weight  0.989995  0.98999 result 1.52423
 training batch 41 mu var0-0.548324
compute loss for weight  0.989985  0.98999 result 1.52422
   --dy = 1.16004 dy_ref = 1.16004
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m1.58567e-10[NON-XML-CHAR-0x1B][39m