12:45 < kris___> https://gist.github.com/kris-singh/ee57aee957c9e1780e982342d1e5bd96

12:46 < kris___> I am getting nearly 100% accuracy on the dataset right now, So i think something must be wrong I will have to check.

13:04 < lozhnikov> kris___: If I remember right "modelssRBM.SampleHidden(std::move(testData.col(i * numPatches + k)), std::move(hiddenMeanTest.slice(i).col(j)));" doesn't work

13:04 < lozhnikov> You should write

13:04 < lozhnikov> arma::mat tmp;

13:04 < lozhnikov> modelssRBM.SampleHidden(std::move(testData.col(i * numPatches + k)), std::move(tmp));

13:04 < lozhnikov> hiddenMeanTest.slice(i).col(j) = tmp;

13:04 < lozhnikov> instead

13:05 < lozhnikov> Look at the classification test. If I remember right I added a workaround for that

16:02 < travis-ci> mlpack/mlpack#3307 (master - 676d5df : Ryan Curtin): The build was broken.

16:02 < travis-ci> Change view : https://github.com/mlpack/mlpack/compare/5729710c7f8a...676d5df25f14

16:02 < travis-ci> Build details : https://travis-ci.org/mlpack/mlpack/builds/270866199

17:51 < kris___> Okay i updated the gist. https://gist.github.com/kris-singh/ee57aee957c9e1780e982342d1e5bd96

17:51 < kris___> But now also i am still getting incorrect results.

17:52 < kris___> Also the classify function in softmax regression should return labels right. Or does it return probabilities.

17:57 < kris___> Btw i am classifying on the spike variables.

18:02 < lozhnikov> kris___: okay, I'll dig into that tomorrow

18:03 < kris___> Okay, right now i was just using random patches for testing as well i just wanted to see how they perfomed.

18:03 < kris___> *performed.

19:54 < kris___> Finally got it working the accuracy is still pretty low at around 14%