00:01 < yashwants19[m]> [zoq](https://matrix.to/#/@freenode_zoq:matrix.org) can you also configure azure pipeline?

00:02 < yashwants19[m]> As it is also running for branch push.

00:03 < yashwants19[m]> I guess we must run pipeline for master only.

00:04 < zoq> I'm not sure I get what you mean.

00:04 < zoq> Right now azure pipeline runs for master and PR's.

00:05 < zoq> https://github.com/mlpack/mlpack/blob/master/.ci/ci.yaml#L1-L8

00:05 < zoq> Or I guess I should say for each branch.

00:07 < yashwants19[m]> I guess azure pipeline is running for push(mlpack/master and mlpack/branchs), pull_request on master and release.

00:07 < zoq> Yes

00:08 < yashwants19[m]> I guess we must turn off the pipeline for push mlpack/branch.

00:08 < yashwants19[m]> Or this is fine.

00:09 < zoq> To solve the mail problem?

00:10 < yashwants19[m]> For these redundant runs.

00:10 < yashwants19[m]> https://dev.azure.com/mlpack/mlpack/_build?definitionId=1

00:11 < zoq> yashwants19[m]: Ahh, I mean it doesn't really matter that much, but we could configure it to run on master and PR's.

00:12 < zoq> It's only twice a month right?

00:12 < yashwants19[m]> Yes

00:12 < yashwants19[m]> 1st and 17th at 10:00 UTC

00:13 < yashwants19[m]> Today I manually run those workflows.

00:14 < zoq> If someone likes to open a PR to fix this, happy to review :)

00:15 < zoq> Right, the schedule should work just fine, fingers crossed.

00:15 < yashwants19[m]> Same here.

00:15 < yashwants19[m]> Fingers crossed 🤞

08:17 < NippunSharmaGitt> Hey, after building mlpack from source (in windows 10) I successfully completed [this](https://www.mlpack.org/doc/mlpack-git/doxygen/sample_ml_app.html) tutorial. Then I tried to make my own application in c++ based on the tutorial (but using linear regression instead of random forest). During that process I encountered these errors:

08:17 < NippunSharmaGitt> error LNK2019: unresolved external symbol "bool __cdecl mlpack::data::Load<double>(class std::basic_string<char,struct std::char_traits<char>,class std::allocator<char> > const &,class arma::Mat<double> &,bool,bool,enum arma::file_type)" (??$Load@N@data@mlpack@@YA_NAEBV?$basic_string@DU?$char_traits@D@std@@V?$allocator@D@2@@std@@AEAV?$Mat@N@arma@@_N2W4file_type@5@@Z) referenced in function main

08:17 < NippunSharmaGitt> error LNK2019: unresolved external symbol "public: double __cdecl mlpack::regression::LinearRegression::Train(class arma::Mat<double> const &,class arma::Row<double> const &,bool)" (?Train@LinearRegression@regression@mlpack@@QEAANAEBV?$Mat@N@arma@@AEBV?$Row@N@5@_N@Z) referenced in function main

08:17 < NippunSharmaGitt> Can someone help here ?

08:18 < NippunSharmaGitt> 'error LNK2019: unresolved external symbol "public: double __cdecl mlpack::regression::LinearRegression::Train(class arma::Mat<double> const &,class arma::Row<double> const &,bool)" (?Train@LinearRegression@regression@mlpack@@QEAANAEBV?$Mat@N@arma@@AEBV?$Row@N@5@_N@Z) referenced in function main'

08:19 < NippunSharmaGitt> `error LNK2019: unresolved external symbol "public: double __cdecl mlpack::regression::LinearRegression::Train(class arma::Mat<double> const &,class arma::Row<double> const &,bool)" (?Train@LinearRegression@regression@mlpack@@QEAANAEBV?$Mat@N@arma@@AEBV?$Row@N@5@_N@Z) referenced in function main`

08:19 < NippunSharmaGitt> `error LNK2019: unresolved external symbol "bool __cdecl mlpack::data::Load<double>(class std::basic_string<char,struct std::char_traits<char>,class std::allocator<char> > const &,class arma::Mat<double> &,bool,bool,enum arma::file_type)" (??$Load@N@data@mlpack@@YA_NAEBV?$basic_string@DU?$char_traits@D@std@@V?$allocator@D@2@@std@@AEAV?$Mat@N@arma@@_N2W4file_type@5@@Z) referenced in function main`

08:19 < NippunSharmaGitt> can someone help here ?

11:42 < SeverinoTessarin> I have a 4GB CSV and I am trying to split into the usual test and training files with ml_preprocess_split but I get the following error message:

11:42 < SeverinoTessarin> error: arma::memory::acquire(): out of memory

11:44 < SeverinoTessarin> /c/mlpack> mlpack_preprocess_split.exe --version

11:44 < SeverinoTessarin> Split Data: part of mlpack git-a8af488.

11:45 < SeverinoTessarin> The MLpack executables are limited by the RAM?

11:51 < zoq> SeverinoTessarin: To use split, your data has to fit into memory, twice. The original and the splitted one.

11:52 < zoq> SeverinoTessarin: Maybe you can increase SWAP?

11:53 < SeverinoTessarin> In Windows?

11:54 < zoq> I think in windows you can increase virtual memory as well, but not sure.

11:55 < SeverinoTessarin> Huge disappointment tho

11:57 < zoq> SeverinoTessarin: Another option would be to do the split inplace.

11:57 < zoq> SeverinoTessarin: So shuffel your data, with arma::shuffel(data);

11:58 < zoq> SeverinoTessarin: And pass the subset to the method with data.cols(0, trainSize - 1); to the method.

11:59 < SeverinoTessarin> Yes, but I would prefer to use the executables API.

12:02 < zoq> I see, so ideally you like to create the train/test set and free the input set at the same time.

12:04 < SeverinoTessarin> if possible, yes.

12:06 < zoq> that is an interesting challenge, and we don't have support for that, but if you like you can open an issue, and I guess someone will pick it up.

12:07 < SeverinoTessarin> Do you know if this limitation applies to the decision tree classifier as well?

12:09 < zoq> All methods take data per reference, so you don't have to keep the data in memory twice.

12:11 < SeverinoTessarin> I will try to open an issue on Github.

12:13 < zoq> Sounds good, thanks.