Sporadic test failures

[aligusnet]

Sometimes some is the test cab fail due to numerical issues. In most case (like 98%) they run just fine.

e.g.:

mltool $ ./scripts/test_runner.py 
mltool-0.1.0.0: test (suite: mltool-test)


running unit tests using command: /Users/ai/src/mltool/.stack-work/dist/x86_64-osx/Cabal-1.24.0.0/build/mltool-test/mltool-test --jxml=mltool-test.xml 
 [============================================================] 100.0%
MachineLearning.SoftmaxClassifier.learn-Conjugate Gradient FR 2
MachineLearning.SoftmaxClassifier.learn-Conjugate Gradient PR 1
MachineLearning.NeuralNetwork.learn-ReLU - Softmax: BFGS 1
MachineLearning.MultiSvmClassifier.learn-BFGS 1
MachineLearning.Classification.OneVsAll.optPath-Gradient Descent 1

[aligusnet]

no test failures any more, will keep it open for a while to retest

mltool $ ./scripts/test_runner.py -n 1000
mltool-0.1.0.2: test (suite: mltool-test)


running unit tests using command: /Users/ai/src/mltool/.stack-work/dist/x86_64-osx/Cabal-1.24.2.0/build/mltool-test/mltool-test --jxml=mltool-test.xml 
 [============================================================] 100.0%
mltool $