Skip to content

Instantly share code, notes, and snippets.

@eraly
Created August 9, 2016 06:34
Show Gist options
  • Save eraly/66844ace9da4d21b8488bd3604dad3af to your computer and use it in GitHub Desktop.
Save eraly/66844ace9da4d21b8488bd3604dad3af to your computer and use it in GitHub Desktop.
23:33:22,221 INFO ~ Running org.nd4j.linalg.lossfunctions.LossFunctionGradientChecks on backend org.nd4j.linalg.cpu.nativecpu.CpuBackend
23:33:24,988 INFO ~ Starting test: LossMSE(), softmax, input shape = [1, 3]
23:33:25,045 INFO ~ Param 2 passed: grad= -0.4444475512228607, numericalGrad= -0.44444444446956055, relError= 6.990194924998631E-4, scorePlus=0.6666569018528489, scoreMinus= 0.6666577907417378
23:33:25,046 INFO ~ Param 2 passed: grad= 0.22221159632756035, numericalGrad= 0.2222147030273014, relError= 0.0013980621888300135, scorePlus=0.6666575685121076, scoreMinus= 0.6666571240827015
23:33:25,049 INFO ~ Param 2 passed: grad= 0.22222663462210948, numericalGrad= 0.22222974144225915, relError= 0.00139802176320326, scorePlus=0.666657568527146, scoreMinus= 0.6666571240676631
DONE
23:33:25,049 INFO ~ Starting test: LossMSE(), softmax, input shape = [4, 4]
23:33:25,067 INFO ~ Param 3 FAILED: grad= -0.07031304072739643, numericalGrad= -0.0937507209641808, relErrorPerc= 24.999999995455152, scorePlus=0.750020716631093, scoreMinus= 0.750020904132535
23:33:25,072 INFO ~ Param 3 FAILED: grad= 0.023435291900536902, numericalGrad= 0.031247188059246866, relErrorPerc= 25.00031728902473, scorePlus=0.7500208416290333, scoreMinus= 0.7500207791346571
23:33:25,075 INFO ~ Param 3 FAILED: grad= 0.023436530125686326, numericalGrad= 0.031249169141212008, relErrorPerc= 25.00110956621314, scorePlus=0.7500208416310143, scoreMinus= 0.7500207791326761
23:33:25,077 INFO ~ Param 3 FAILED: grad= 0.023439776783114184, numericalGrad= 0.03125436387474423, relErrorPerc= 25.003187148354645, scorePlus=0.7500208416362091, scoreMinus= 0.7500207791274813
23:33:25,080 INFO ~ Param 3 FAILED: grad= 0.02343128852144433, numericalGrad= 0.031240247166941515, relErrorPerc= 24.996468829992608, scorePlus=0.7500208416220924, scoreMinus= 0.750020779141598
23:33:25,081 INFO ~ Param 3 FAILED: grad= -0.07031263922870527, numericalGrad= -0.09375018605872754, relErrorPerc= 25.000000336362405, scorePlus=0.750020716631628, scoreMinus= 0.7500209041320001
23:33:25,084 INFO ~ Param 3 FAILED: grad= 0.023436448981064024, numericalGrad= 0.031248503451486442, relErrorPerc= 24.999771533221413, scorePlus=0.7500208416303487, scoreMinus= 0.7500207791333418
23:33:25,089 INFO ~ Param 3 FAILED: grad= 0.023444530903630512, numericalGrad= 0.031261435329277276, relErrorPerc= 25.004944089454515, scorePlus=0.7500208416432805, scoreMinus= 0.7500207791204099
23:33:25,091 INFO ~ Param 3 FAILED: grad= 0.02344059503879486, numericalGrad= 0.03125386982549827, relErrorPerc= 24.999383533392074, scorePlus=0.750020841635715, scoreMinus= 0.7500207791279754
23:33:25,097 INFO ~ Param 3 FAILED: grad= 0.023440446861696728, numericalGrad= 0.03125363268186021, relErrorPerc= 24.9992885617367, scorePlus=0.7500208416354779, scoreMinus= 0.7500207791282125
23:33:25,101 INFO ~ Param 3 FAILED: grad= -0.07031168834585709, numericalGrad= -0.09374891807301111, relErrorPerc= 25.000000222830565, scorePlus=0.7500207166328958, scoreMinus= 0.750020904130732
23:33:25,106 INFO ~ Param 3 FAILED: grad= 0.023432810984433252, numericalGrad= 0.03124141556565263, relErrorPerc= 24.994400669233112, scorePlus=0.7500208416232608, scoreMinus= 0.7500207791404296
23:33:25,108 INFO ~ Param 3 FAILED: grad= 0.023441389764184983, numericalGrad= 0.03125119635294382, relErrorPerc= 24.990424368259955, scorePlus=0.7500208416330415, scoreMinus= 0.7500207791306488
23:33:25,114 INFO ~ Param 3 FAILED: grad= 0.023434927113361453, numericalGrad= 0.03124085690142664, relErrorPerc= 24.986285788174783, scorePlus=0.7500208416227021, scoreMinus= 0.7500207791409883
23:33:25,117 INFO ~ Param 3 FAILED: grad= 0.023442466490014947, numericalGrad= 0.03125291903049998, relErrorPerc= 24.99111373520902, scorePlus=0.7500208416347642, scoreMinus= 0.7500207791289262
23:33:25,118 INFO ~ Param 3 FAILED: grad= -0.07030872911137294, numericalGrad= -0.0937449723403816, relErrorPerc= 25.000000153515703, scorePlus=0.7500207166368417, scoreMinus= 0.7500209041267863
DONE
23:33:25,121 INFO ~ Ending org.nd4j.linalg.lossfunctions.LossFunctionGradientChecks
Process finished with exit code 0
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment