04:39 < favre49> Could someone explain what this line does to me? I don't really understand the syntax that's being used, I'm not that well acquainted with numpy. https://github.com/msu-coinlab/pymoo/blob/master/pymoo/algorithms/nsga3.py#L166

07:51 < jenkins-mlpack2> Project docker mlpack nightly build build #407: STILL UNSTABLE in 3 hr 37 min: http://ci.mlpack.org/job/docker%20mlpack%20nightly%20build/407/

10:07 < KimSangYeon-DGU> favre49: This is my brief understanding the numpy code that you linked

10:08 < KimSangYeon-DGU> favre49: At line 153, `asf` is an identity matrix with 2 dimension (column of F, column of F)

10:08 < KimSangYeon-DGU> At line 154, non diagonal element of 'asf' will be replaced by 1e6

10:08 < KimSangYeon-DGU> At line 166, 'None' means 'Add new axis with 1', so asf[:,None,:] means that create a new matrix with the 3 dimension (column of F, 1, column of F).

10:08 < KimSangYeon-DGU> About the np.max(), for example, if there is a 3D matrix A = [[[1, 2, 3]], [[5, 1, 0]], [[0, 0, 1]]], then the np.max(A, axis=2) is [[3], [5], [1]]. The parameter `axis=2` means computation with keeping row and column dimension.

14:59 < jeffin143> lozhnikov : I cannot pass policy.Encode(output, dictionary.Value(token), i, numTokens); for tfidf

14:59 < jeffin143> Since i have to calculate tfidf values and hence I have to pass the output matrix as a whole

15:00 < jeffin143> and hence I cannot call the policy.Encode inside the loop and have to call it after the loop ends

15:01 < lozhnikov> jeffin143: I think the policy itself should calculate the values.

15:02 < jeffin143> https://github.com/jeffin143/mlpack/blob/fafb6c330778d8feb20ac3baceda6e864183640b/src/mlpack/core/data/tfidf_encoding_impl.hpp#L110-L138

15:03 < jeffin143> Can you take a look at, I need to get the count, but the count will only be calculated after iterating over the token once for a particular

15:03 < jeffin143> row

15:05 < lozhnikov> Yes, right now you iterate over the tokens twice. You can calculate the amount of tokens during the first pass. Of course you should introduce a policy function for that.

15:06 < lozhnikov> The idfdict variable should be defined inside the policy.

15:07 < jeffin143> Ok then I need to change the StringEncoding Class , ryt ?

15:10 < lozhnikov> I think it's enough to add a call to the policy function into the first pass of EncodeHelper(). The function does nothing in case of other policies. You don't need to change anything else.

15:21 < jeffin143> ok, I will try

15:58 < jeffin143> lozhnikov : how to pass dictionary of StringEncoding

15:58 < jeffin143> inside a function

15:58 < jeffin143> MatType,typename DictionaryType> static void En(MatType& output, DictionaryType& dictionary) { /// }

15:58 < jeffin143> template<typename MatType,typename DictionaryType> static void En(MatType& output, DictionaryType& dictionary) { // }

16:03 < lozhnikov> is En() a policy function?

16:05 < lozhnikov> Yes, I think it's correct. But I am not quite sure that you need the whole dictionary.

16:07 < jeffin143> i just need the mapping

16:07 < jeffin143> right ?

16:07 < jeffin143> Yes En is a policy function, I was trying it out, It doesn't get pass that way

16:08 < lozhnikov> I thought it's enough to call the function for each token. In that case you don

16:08 < lozhnikov> In that case you don't need to pass the whole dictionary.

16:19 < lozhnikov> jeffin143: I mean something like this: https://pastebin.com/RSN4WCAh

16:23 < jeffin143> I am story the intermediate results inside output matrix

16:24 < jeffin143> and hence I need the output matrix to be build before the preprocess thing

16:24 < jeffin143> i mean initMatrix !

16:27 < lozhnikov> Your approach requires three passes. I am not sure it's reasonable. Let's implement my approach. Then I'll benchmark them.

16:31 < jeffin143> Then i need two data structure, one to store idfvalues and the other one to store the count of words

16:32 < lozhnikov> jeffin143: Yes, I think it's reasonable. Look at the example.

20:30 < jenkins-mlpack2> Project mlpack - git commit test build #208: STILL UNSTABLE in 47 min: http://ci.mlpack.org/job/mlpack%20-%20git%20commit%20test/208/

20:46 < travis-ci> mlpack/mlpack#7680 (master - 12a50a0 : Ryan Curtin): The build was fixed.

20:46 < travis-ci> Change view : https://github.com/mlpack/mlpack/compare/a8dd9f530101...12a50a055ba7

20:46 < travis-ci> Build details : https://travis-ci.org/mlpack/mlpack/builds/567617650