Execution Time0.12s

Test: TMVA-DNN-BatchNormalization (Passed)
Build: master-aarch64-centos7-gcc48 (techlab-arm64-moonshot-xgene-003) on 2019-11-13 00:50:30
Repository revision: 30660dce2d9e89e4852dbf83dbd8b2cfcc137eff

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 var00.103594
output DL 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    0.04116      0.7769 
   1 |      1.239       1.004 
   2 |     -0.934       1.195 
   3 |     0.1611       2.641 
   4 |    -0.1009       1.236 
   5 |     0.2786       2.376 
   6 |    -0.6611      -2.551 
   7 |      1.005     -0.1888 
   8 |    0.00551     -0.9169 
   9 |   0.001135      0.1928 

output BN 
output DL feature 0 mean 0.103594	output DL std 0.655212
output DL feature 1 mean 0.576456	output DL std 1.53799
output of BN 

10x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.1004      0.1374 
   1 |      1.827      0.2931 
   2 |     -1.669      0.4238 
   3 |    0.09258       1.415 
   4 |     -0.329       0.452 
   5 |     0.2815       1.233 
   6 |      -1.23      -2.144 
   7 |       1.45     -0.5245 
   8 |    -0.1578      -1.024 
   9 |    -0.1648     -0.2629 

output BN feature 0 mean 5.82867e-17	output BN std 1.05396
output BN feature 1 mean 5.55112e-17	output BN std 1.05407
Testing weight gradients   for    layer 0
weight gradient for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |       0.54      -1.524      0.1265      -2.333 
   1 |    0.02146      0.7999     -0.4784      0.3574 

weights for layer 0

2x4 matrix is as follows

     |      0    |      1    |      2    |      3    |
---------------------------------------------------------
   0 |     0.7965      0.5563      -0.664     -0.2152 
   1 |     0.7978     -0.2994     -0.2471      -1.431 

 training batch 2 mu var00.103597
compute loss for weight  0.796556  0.796546 result 2.65228
 training batch 3 mu var00.103594
compute loss for weight  0.796536  0.796546 result 2.65227
 training batch 4 mu var00.103595
compute loss for weight  0.796551  0.796546 result 2.65227
 training batch 5 mu var00.103594
compute loss for weight  0.796541  0.796546 result 2.65227
   --dy = 0.540007 dy_ref = 0.540007
 training batch 6 mu var00.103594
compute loss for weight  0.556337  0.556327 result 2.65226
 training batch 7 mu var00.103594
compute loss for weight  0.556317  0.556327 result 2.65229
 training batch 8 mu var00.103594
compute loss for weight  0.556332  0.556327 result 2.65226
 training batch 9 mu var00.103594
compute loss for weight  0.556322  0.556327 result 2.65228
   --dy = -1.52416 dy_ref = -1.52416
 training batch 10 mu var00.103594
compute loss for weight  -0.664017  -0.664027 result 2.65227
 training batch 11 mu var00.103594
compute loss for weight  -0.664037  -0.664027 result 2.65227
 training batch 12 mu var00.103594
compute loss for weight  -0.664022  -0.664027 result 2.65227
 training batch 13 mu var00.103594
compute loss for weight  -0.664032  -0.664027 result 2.65227
   --dy = 0.126535 dy_ref = 0.126535
 training batch 14 mu var00.103594
compute loss for weight  -0.215206  -0.215216 result 2.65225
 training batch 15 mu var00.103594
compute loss for weight  -0.215226  -0.215216 result 2.6523
 training batch 16 mu var00.103594
compute loss for weight  -0.215211  -0.215216 result 2.65226
 training batch 17 mu var00.103594
compute loss for weight  -0.215221  -0.215216 result 2.65228
   --dy = -2.3334 dy_ref = -2.3334
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |      1.445       3.859 

weights for layer 1

1x2 matrix is as follows

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

 training batch 18 mu var00.103594
compute loss for weight  1.00001  1 result 2.65229
 training batch 19 mu var00.103594
compute loss for weight  0.99999  1 result 2.65226
 training batch 20 mu var00.103594
compute loss for weight  1.00001  1 result 2.65228
 training batch 21 mu var00.103594
compute loss for weight  0.999995  1 result 2.65226
   --dy = 1.44521 dy_ref = 1.44521
 training batch 22 mu var00.103594
compute loss for weight  1.00001  1 result 2.65231
 training batch 23 mu var00.103594
compute loss for weight  0.99999  1 result 2.65223
 training batch 24 mu var00.103594
compute loss for weight  1.00001  1 result 2.65229
 training batch 25 mu var00.103594
compute loss for weight  0.999995  1 result 2.65225
   --dy = 3.85934 dy_ref = 3.85934
Testing weight gradients   for    layer 1
weight gradient for layer 1

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |  6.939e-17   3.192e-16 

weights for layer 1

1x2 matrix is as follows

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

 training batch 26 mu var00.103594
compute loss for weight  1e-05  0 result 2.65227
 training batch 27 mu var00.103594
compute loss for weight  -1e-05  0 result 2.65227
 training batch 28 mu var00.103594
compute loss for weight  5e-06  0 result 2.65227
 training batch 29 mu var00.103594
compute loss for weight  -5e-06  0 result 2.65227
   --dy = 5.92119e-11 dy_ref = 6.93889e-17
 training batch 30 mu var00.103594
compute loss for weight  1e-05  0 result 2.65227
 training batch 31 mu var00.103594
compute loss for weight  -1e-05  0 result 2.65227
 training batch 32 mu var00.103594
compute loss for weight  5e-06  0 result 2.65227
 training batch 33 mu var00.103594
compute loss for weight  -5e-06  0 result 2.65227
   --dy = 7.40149e-12 dy_ref = 3.19189e-16
Testing weight gradients   for    layer 2
weight gradient for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |     -2.017      -2.942 

weights for layer 2

1x2 matrix is as follows

     |      0    |      1    |
-------------------------------
   0 |    -0.7167      -1.312 

 training batch 34 mu var00.103594
compute loss for weight  -0.716646  -0.716656 result 2.65225
 training batch 35 mu var00.103594
compute loss for weight  -0.716666  -0.716656 result 2.65229
 training batch 36 mu var00.103594
compute loss for weight  -0.716651  -0.716656 result 2.65226
 training batch 37 mu var00.103594
compute loss for weight  -0.716661  -0.716656 result 2.65228
   --dy = -2.0166 dy_ref = -2.0166
 training batch 38 mu var00.103594
compute loss for weight  -1.31171  -1.31172 result 2.65224
 training batch 39 mu var00.103594
compute loss for weight  -1.31173  -1.31172 result 2.6523
 training batch 40 mu var00.103594
compute loss for weight  -1.31171  -1.31172 result 2.65226
 training batch 41 mu var00.103594
compute loss for weight  -1.31172  -1.31172 result 2.65229
   --dy = -2.9422 dy_ref = -2.9422
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][32m6.45897e-10[NON-XML-CHAR-0x1B][39m