Execution Time0.07s

Test: TMVA-DNN-BatchNormalization-Cpu (Passed)
Build: master-x86_64-fedora28-gcc8 (sft-fedora-28-1.cern.ch) on 2019-11-14 01:14:20

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 
 training batch 1 mu var0-0.382054
output DL 
output BN 
output DL feature 0 mean -0.382054	output DL std 1.29841
output DL feature 1 mean -0.687013	output DL std 1.35136
output of BN 
output BN feature 0 mean 2.77556e-17	output BN std 1.05406
output BN feature 1 mean -2.22045e-17	output BN std 1.05406
Testing weight gradients   for    layer 0
weight gradient for layer 0
weights for layer 0
 training batch 2 mu var0-0.382052
compute loss for weight  -0.146191  -0.146201 result 2.70241
 training batch 3 mu var0-0.382054
compute loss for weight  -0.146211  -0.146201 result 2.70241
 training batch 4 mu var0-0.382054
compute loss for weight  -0.146196  -0.146201 result 2.70241
 training batch 5 mu var0-0.382054
compute loss for weight  -0.146206  -0.146201 result 2.70241
   --dy = -0.149365 dy_ref = -0.149365
 training batch 6 mu var0-0.382055
compute loss for weight  1.34905  1.34904 result 2.70241
 training batch 7 mu var0-0.382054
compute loss for weight  1.34903  1.34904 result 2.70241
 training batch 8 mu var0-0.382055
compute loss for weight  1.34904  1.34904 result 2.70241
 training batch 9 mu var0-0.382054
compute loss for weight  1.34903  1.34904 result 2.70241
   --dy = -0.00498669 dy_ref = -0.00498669
 training batch 10 mu var0-0.382054
compute loss for weight  0.265129  0.265119 result 2.70241
 training batch 11 mu var0-0.382054
compute loss for weight  0.265109  0.265119 result 2.70241
 training batch 12 mu var0-0.382054
compute loss for weight  0.265124  0.265119 result 2.70241
 training batch 13 mu var0-0.382054
compute loss for weight  0.265114  0.265119 result 2.70241
   --dy = -0.0496577 dy_ref = -0.0496577
 training batch 14 mu var0-0.382054
compute loss for weight  -0.188669  -0.188679 result 2.70241
 training batch 15 mu var0-0.382054
compute loss for weight  -0.188689  -0.188679 result 2.70241
 training batch 16 mu var0-0.382054
compute loss for weight  -0.188674  -0.188679 result 2.70241
 training batch 17 mu var0-0.382054
compute loss for weight  -0.188684  -0.188679 result 2.70241
   --dy = 0.00839415 dy_ref = 0.00839415
Testing weight gradients   for    layer 1
weight gradient for layer 1
weights for layer 1
 training batch 18 mu var0-0.382054
compute loss for weight  1.00001  1 result 2.70247
 training batch 19 mu var0-0.382054
compute loss for weight  0.99999  1 result 2.70236
 training batch 20 mu var0-0.382054
compute loss for weight  1.00001  1 result 2.70244
 training batch 21 mu var0-0.382054
compute loss for weight  0.999995  1 result 2.70238
   --dy = 5.4778 dy_ref = 5.4778
 training batch 22 mu var0-0.382054
compute loss for weight  1.00001  1 result 2.70241
 training batch 23 mu var0-0.382054
compute loss for weight  0.99999  1 result 2.70241
 training batch 24 mu var0-0.382054
compute loss for weight  1.00001  1 result 2.70241
 training batch 25 mu var0-0.382054
compute loss for weight  0.999995  1 result 2.70241
   --dy = -0.0729762 dy_ref = -0.0729762
Testing weight gradients   for    layer 1
weight gradient for layer 1
weights for layer 1
 training batch 26 mu var0-0.382054
compute loss for weight  1e-05  0 result 2.70241
 training batch 27 mu var0-0.382054
compute loss for weight  -1e-05  0 result 2.70241
 training batch 28 mu var0-0.382054
compute loss for weight  5e-06  0 result 2.70241
 training batch 29 mu var0-0.382054
compute loss for weight  -5e-06  0 result 2.70241
   --dy = -1.11022e-10 dy_ref = 1.45717e-16
 training batch 30 mu var0-0.382054
compute loss for weight  1e-05  0 result 2.70241
 training batch 31 mu var0-0.382054
compute loss for weight  -1e-05  0 result 2.70241
 training batch 32 mu var0-0.382054
compute loss for weight  5e-06  0 result 2.70241
 training batch 33 mu var0-0.382054
compute loss for weight  -5e-06  0 result 2.70241
   --dy = 1.25825e-10 dy_ref = -1.51788e-17
Testing weight gradients   for    layer 2
weight gradient for layer 2
weights for layer 2
 training batch 34 mu var0-0.382054
compute loss for weight  -1.66833  -1.66834 result 2.70238
 training batch 35 mu var0-0.382054
compute loss for weight  -1.66835  -1.66834 result 2.70244
 training batch 36 mu var0-0.382054
compute loss for weight  -1.66833  -1.66834 result 2.70239
 training batch 37 mu var0-0.382054
compute loss for weight  -1.66834  -1.66834 result 2.70243
   --dy = -3.28339 dy_ref = -3.28339
 training batch 38 mu var0-0.382054
compute loss for weight  -0.0881488  -0.0881588 result 2.70242
 training batch 39 mu var0-0.382054
compute loss for weight  -0.0881688  -0.0881588 result 2.7024
 training batch 40 mu var0-0.382054
compute loss for weight  -0.0881538  -0.0881588 result 2.70242
 training batch 41 mu var0-0.382054
compute loss for weight  -0.0881638  -0.0881588 result 2.70241
   --dy = 0.827781 dy_ref = 0.827781
Testing weight gradients:      maximum relative error: [NON-XML-CHAR-0x1B][33m1.17685e-08[NON-XML-CHAR-0x1B][39m