IRC logs of #shogun for Monday, 2016-06-06

« Sunday, 2016-06-05 Index Tuesday, 2016-06-07 »
--- Log opened Mon Jun 06 00:00:28 2016
-!- GandalfTheWizar1 [~Eva@112.10.170.58] has quit [Ping timeout: 240 seconds]00:04
-!- GandalfTheWizard [~Eva@112.10.170.58] has joined #shogun00:05
-!- sonne|osx [~sonne@x4db3bf36.dyn.telefonica.de] has quit [Quit: sonne|osx]00:45
-!- lambday [6d9941c0@gateway/web/freenode/ip.109.153.65.192] has quit [Ping timeout: 250 seconds]01:44
-!- arianepaola [~ariane@unaffiliated/arianepaola] has quit [Read error: Connection reset by peer]04:53
-!- arianepaola [~ariane@unaffiliated/arianepaola] has joined #shogun04:58
-!- sonne|osx [~sonne@x4db3bf36.dyn.telefonica.de] has joined #shogun05:41
-!- sonne|osx [~sonne@x4db3bf36.dyn.telefonica.de] has quit [Quit: sonne|osx]05:57
-!- sanuj [~sanuj@117.203.23.253] has joined #shogun07:04
sanujSaurabh7, hey07:33
@wikinghi07:35
@wikingcan i get a standup from both of you07:35
sanujwiking, hi07:35
@wikingsanuj & Saurabh707:35
sanujwiking, My PR for tags is merge ready (reviewed by sergey) https://github.com/shogun-toolbox/shogun/pull/322107:36
sanujwas working on this last week07:36
sanujwiking, lisitsyn  insists that you or HeikoS review it once before we merge it07:37
@wikingk07:37
@wikingfirst of all you should stash your commits into 107:37
sanujwiking, there are lots of changes, you want one commit for that?07:37
@wikingyes07:38
sanujwiking, neural net cookbook https://github.com/shogun-toolbox/shogun/pull/324607:38
@wikings/stash/squash/g07:38
@wikinghttps://ariejan.net/2011/07/05/git-squash-your-latests-commits-into-one/07:38
sanuji was having some problems with the cookbook07:38
sanujthanks for the link07:38
sanuji'm making a cookbook for mkl also07:41
sanujthis was due last week but i got stuck with the nn cookbook07:41
Saurabh7hello07:51
Saurabh7sanuj: hey07:51
sanujSaurabh7, you wrote the mkl notebook?07:51
sanuji have some kernel doubts :)07:51
Saurabh7wiking: finsihed multithread support for unlcoked xval, working on polishing LARS and eigenising it, done for some important routines and added unit tests07:52
Saurabh7also fixed lars benchmark in the frameowrk07:53
Saurabh7sanuj: yep:)07:53
Saurabh7sanuj: whats the problem?07:53
sanujSaurabh7, so via kernels we are trying to project the data in a different space right?07:53
Saurabh7sanuj: yes07:54
sanujso that it is linearly separable in the new space07:54
Saurabh7sanuj: yes tahts the idea07:54
sanujSaurabh7, every kernel takes 2 features as argument07:54
sanujwhy is that?07:54
Saurabh7sanuj: its basically jsut like distance class07:55
Saurabh7kernel function si computed for kernel(i,j)07:55
Saurabh7i is lhs and j rhs features index07:55
sanujSaurabh7, what do you mean by lhs and rhs features07:56
sanuji saw this in the doc ^07:56
sanujbut it confused me07:56
Saurabh7sanuj: you need two points for calucalting say euclidean dist07:58
Saurabh7same way for kernel07:58
Saurabh7they can come from two different features07:58
Saurabh7matrices07:58
Saurabh7kernel fucntion between ith sample of lhs and jth sample of rhs07:59
sanujoh i see07:59
sanujSaurabh7, if you look at this https://github.com/shogun-toolbox/shogun/blob/develop/examples/undocumented/python_modular/mkl_binclass_modular.py08:01
sanujSaurabh7, so while training we keep lhs = rhs = training_features08:02
sanujand while testing we have lhs = training_features and rhs = testing_features08:03
Saurabh7sanuj: yes08:03
sanujSaurabh7, if we are generating the kernel matrices like this, then what does mkl.train() learns?08:04
sanujthe weights of linear combination of kernels?08:05
Saurabh7sanuj: yes optimal weights08:05
sanujokay08:06
sanujSaurabh7, thanks :)08:06
Saurabh7sanuj: cool :)08:06
-!- sanuj [~sanuj@117.203.23.253] has quit [Ping timeout: 240 seconds]09:02
-!- sanuj [~sanuj@117.203.23.253] has joined #shogun09:42
-!- c4goldsw [c1a99ae1@gateway/web/cgi-irc/kiwiirc.com/ip.193.169.154.225] has joined #shogun10:42
c4goldswHey, does anyone know if SGVector supports scalar multiplication?  I'm browsing through the corresponding .h and .cpp file and I can't find anything, I just want to be certain.10:55
@wikingc4goldsw: scale_vector10:56
@wikingor10:56
@wiking/// Scale vector inplace10:56
@wikingvoid scale(T alpha);10:56
c4goldswwiking thanks!10:58
-!- c4goldsw [c1a99ae1@gateway/web/cgi-irc/kiwiirc.com/ip.193.169.154.225] has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client]11:01
-!- c4goldsw [c1a99ae1@gateway/web/cgi-irc/kiwiirc.com/ip.193.169.154.225] has joined #shogun11:14
-!- lambday [8028b10a@gateway/web/freenode/ip.128.40.177.10] has joined #shogun11:36
-!- mode/#shogun [+o lambday] by ChanServ11:36
-!- c4goldsw [c1a99ae1@gateway/web/cgi-irc/kiwiirc.com/ip.193.169.154.225] has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client]12:11
-!- c4goldsw [c1a99ae1@gateway/web/cgi-irc/kiwiirc.com/ip.193.169.154.225] has joined #shogun12:11
-!- c4goldsw [c1a99ae1@gateway/web/cgi-irc/kiwiirc.com/ip.193.169.154.225] has quit [Client Quit]12:15
-!- besser82 [~besser82@fedora/besser82] has joined #shogun12:18
-!- mode/#shogun [+o besser82] by ChanServ12:18
sanujSaurabh7, there?12:28
-!- c4goldsw [c1a99ae1@gateway/web/cgi-irc/kiwiirc.com/ip.193.169.154.225] has joined #shogun13:49
c4goldswlambday, you around?13:51
@lambdayc4goldsw: yeah13:51
c4goldswwould you have time to look at a PR I made?13:51
@lambdayc4goldsw: sure..13:57
c4goldswGreat - it's not too big: https://github.com/shogun-toolbox/shogun/pull/324913:57
@lambdayc4goldsw: oh you added variance methods for linalg13:59
c4goldswYes.13:59
c4goldswlambday: I'm just looking for feedback - my main concern is if I'm using eigen to the full extent that I can to take advantage of SIMD operations.14:00
c4goldswmy main concern is to ensure that I'm using*14:00
@lambdayI see.14:01
@lambdaylet me have a look. But did you have a look at https://github.com/shogun-toolbox/shogun/blob/develop/src/shogun/mathematics/Statistics.h#L60 ?14:01
@lambdayI wonder whether the method you added does the same14:02
@lambdayc4goldsw:14:02
c4goldswIt is - Heiko wanted to move some stats methods out of CStatistics into linalg.  The issue is here: https://github.com/shogun-toolbox/shogun/issues/311814:02
c4goldswlambday:14:03
@lambdayc4goldsw: ah I see14:03
c4goldswI guess this is being done so that these computations can be done with whatver linalg backend a user uses.14:03
@lambdayc4goldsw: okay, so first of all, we're changing the structure of linalg a bit.. so all this linalg code probably be refactored in some way.. this is what OXPHOS is working on for her GSoC project..14:04
@lambdaybut let me comment on the usage of the eigen3 part anyway14:04
-!- sanuj [~sanuj@117.203.23.253] has quit [Ping timeout: 246 seconds]14:04
c4goldswlambday: Thanks!14:05
c4goldswlambday I read / replied to your comment - we can talk over here if that's easier.14:13
c4goldswlambday: now I understand what you said about replacing sqaureResult14:15
-!- sanuj [~sanuj@117.203.23.253] has joined #shogun14:19
-!- sanuj [~sanuj@117.203.23.253] has quit [Client Quit]14:20
-!- sanuj [~sanuj@117.203.23.253] has joined #shogun14:21
lisitsynlambday: hey14:24
@lambdaylisitsyn: hello14:25
lisitsynlambday: could you please review sanuj's code?14:25
@lambdaylisitsyn: which PR you want me to have a look?14:25
sanujlambday, lisitsyn hey14:26
lisitsynin a sec14:26
sanuji'll give you the link14:26
@lambdayc4goldsw: yeah if possible, we can get rid of these temporaries14:26
lisitsynlambday: https://github.com/shogun-toolbox/shogun/pull/322114:26
lisitsynlambday: be super tough as it is critical! :)14:26
sanujyeah this14:26
@lambdaylisitsyn: thanks.. checking14:26
@lambdayhah14:26
sanujlambday, when i heiko coming back?14:26
c4goldswlambday sure - thanks for looking over it.14:26
@lambdaysanuj: should have been back by now14:27
sanujokay14:27
@lambdayshall I also comment on styling etc?14:28
sanujlisitsyn, ^14:28
lisitsynlambday: yes sure14:28
lisitsynlambday: we target to merge that so lets polish that14:29
@lambdayok14:29
c4goldswlambday another quick question: did the last compute() function also check out?  There were no comments.14:29
sanujlisitsyn, wiking told me to squish the commits14:29
sanuji'll do that once there are no more changes to be done14:29
lisitsynsanuj: yes14:30
lisitsynmakes sense14:30
sanujlisitsyn, we will merge this in shogun/tags or shogun/develop?14:30
lisitsynsanuj: I thought about develop14:30
sanujokay, the PR is against shogun/tags14:30
lisitsynsanuj: well probably it makes sense to get something complete in shogun/tags14:40
lisitsynlike with swig14:40
sanujokay14:40
@lambdaysanuj: commented on your PR15:06
@lambdaythere are some minor styling issues15:06
sanujyes the spaces15:06
sanuji'll fix those15:06
@lambdayjust a couple of other things15:06
sanujlambday, thanks :)15:07
@lambdaymaybe the constructors for BaseTag and Tag should be explicit15:07
sanujokay15:07
@lambdayalso, whenever you add operator==, please also add operator!=15:07
@lambdaymost important thing that is missing from the PR is class documentation15:07
@lambdayplease add those.. for each method and class15:08
* lambday afk15:09
sanujlambday, sure15:11
lisitsynlambday: nice catch on !=15:11
@wikingmmm15:13
@wikingoxphos is not around yet?15:13
@wikingarianepaola: how's the setup.py going/15:14
@wiking?15:14
* lambday re15:36
@wikingre15:36
@wikingfe15:36
@lambdayheh15:37
@lambdaylisitsyn: yeah that just caught my attention :D15:37
sanujre15:40
@lambday(R)eal (E)vil15:41
sanuj:P15:42
arianepaolahi wiking, just to let you know that I will be online later today (presentation at uni).15:42
@lambdayc4goldsw: hey.. yeah I had a quick look.. the thing to care about in these methods is that we want to minimize (if possible, avoid) temporary object creation..15:43
c4goldswlambday Okee Dokee15:44
@wikingarianepaola: k15:47
-!- sanuj [~sanuj@117.203.23.253] has quit [Ping timeout: 252 seconds]15:49
-!- c4goldsw [c1a99ae1@gateway/web/cgi-irc/kiwiirc.com/ip.193.169.154.225] has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client]15:52
-!- OXPHOS [9d8b131c@gateway/web/freenode/ip.157.139.19.28] has joined #shogun16:05
@wikingOXPHOS: yo16:05
-!- c4goldsw [c1a99ae1@gateway/web/cgi-irc/kiwiirc.com/ip.193.169.154.225] has joined #shogun16:06
OXPHOSwiking: hi16:08
OXPHOSwiking: log: linalg working prototype. Now works with Eigen3 and ViennaCL. Also linalg benchmark. Cookbook: multi-class logistic regression and multi-class linear machine16:11
OXPHOSlambday: hey16:12
@lambdayOXPHOS: hello16:12
@wikingOXPHOS: should talk about cereal16:12
OXPHOSwiking: you mean now? didn't do much last week16:13
@wikingi know16:17
@wikingbut yeah16:17
@wikingwe should talk16:17
@wikingas that part needs to get rolling as well16:18
-!- HeikoS [~heiko@nat-182-210.internal.eduroam.ucl.ac.uk] has joined #shogun16:21
-!- mode/#shogun [+o HeikoS] by ChanServ16:21
@HeikoSjojojo16:21
OXPHOSwiking: sure! I'll definitely read sanuj's code16:21
@HeikoSSaurabh7:  hi there16:21
@HeikoSOXPHOS:  hi!16:21
OXPHOSHeikoS: hello!16:21
c4goldswHeikoS Hello16:21
@HeikoSc4goldsw: ah hello16:21
@wikingOXPHOS: ? what's that has to do with this? :)16:21
@wikingi mean with cereal16:21
OXPHOSwiking: we're doing it in parameter.h? or u mean sth. more basic16:22
@wikingOXPHOS: so your problem was obvious16:22
@wikingwith sgvector16:23
@wikingwhy did you keep the old load/save vector... because the problem is that since the new one is a template it'll unfortunately match some other code16:23
@wiking*function16:23
@wikingbut16:24
@wikingmaybe we should use a different naming16:26
@wikinglike serialize16:26
@wikingand deserialize16:26
-!- sanuj [~sanuj@117.203.18.75] has joined #shogun16:26
@wikingthat way it wont have that problem16:26
lisitsynhey16:26
lisitsynwiking: OXPHOS: got time to instant chat on what sanuj is doing?16:26
@wikingyesah16:26
@wikingi'm about to pass out16:26
@wiking but tell me16:26
lisitsynok simple words16:26
OXPHOSlisitsyn: sure16:27
@HeikoSwiking: passing out16:27
lisitsynSGObject was base class with nothing16:27
lisitsynfields are added in inherited classes16:27
lisitsynwhat we're changing now is adding some capability16:27
lisitsynto store parameters of any type16:27
lisitsynbasically it is16:27
lisitsynmap<Tag, Any>16:27
lisitsynwe would need to add serialize/deserialize things for that16:28
OXPHOSlisitsyn: so just serealize Tag and any16:29
OXPHOSwiking: did you mean change the name of cereal save/load to deserialize? not sure that can be done16:29
@wikingwhy couldnt' it be done?16:30
@wikingwe could call it 'la' and 'lala'16:30
OXPHOSwiking: overload the cereal header?16:30
@wiking?16:30
@wikingjust a sec16:30
@wiking(meeting)16:30
lisitsynOXPHOS: yes we can obtain name and type from Tag16:31
lisitsynand name and type from Any16:31
OXPHOSlisitsyn: cool16:32
OXPHOSwiking: yeah one can change that in cereal/macros.hpp16:33
sanujHeikoS, you are back!16:33
@HeikoSsanuj: yes, hi16:33
@HeikoSOXPHOS: static calls will work now, I just commented on the kmeans PR16:33
OXPHOSHeikoS: thx!16:34
sanujHeikoS, that means maths:init_rand(0) should work now16:34
sanuji'll rebase my gmm PR16:35
@HeikoSsanuj: yes :)16:35
@HeikoSlets hope these PRs can be merged now16:35
@HeikoSwe fixed two issues in the meta example system16:35
sanujthat's good16:36
shogun-buildbotbuild #2894 of bsd1 - libshogun is complete: Failure [failed configure]  Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/2894  blamelist: Esben Sorig <esben@sorig.eu>, Heiko Strathmann <heiko.strathmann@gmail.com>16:38
@HeikoSsanuj: let me know ok? then we can merge these guys asap16:40
c4goldswHeikoS: Hey, would you have time to look at a PR I've made?16:40
sanujHeikoS, remove examples/meta/lextab.py, examples/meta/parsetab.py?16:40
@HeikoSc4goldsw: sure, will do soon. I was away for a few days and just going through emails16:40
c4goldswHeikoS: Great, thanks.16:40
@HeikoSsanuj: yeah, everytime the meta example grammar is changed, we need to re-create these (happens automatically)16:40
@HeikoSsanuj: I will soon fix this behaviour16:41
sanujokay, cool16:41
shogun-buildbotbuild #32 of xenial - libshogun is complete: Failure [failed compile]  Build details are at http://buildbot.shogun-toolbox.org/builders/xenial%20-%20libshogun/builds/32  blamelist: Esben Sorig <esben@sorig.eu>, Heiko Strathmann <heiko.strathmann@gmail.com>16:41
sanujHeikoS, getting same error while generating meta examples16:58
sanujhttp://pastebin.com/prC4FDdx16:58
sanujdo you think this might be related to python 316:58
sanujyou remember i told you about most of my python modular tests failing16:58
@HeikoScd examples/meta/generator/16:58
@HeikoSrm lextab.py* parser.out parsetab.py*16:58
@HeikoSthis didnt work?16:59
sanuji think that was because of python316:59
@HeikoSno16:59
@HeikoSnot python 316:59
sanuji removed lextab.py16:59
sanujokay let me try what you said16:59
@HeikoSwe have a build where it works with Python 316:59
@HeikoSlet me forward you an email16:59
sanujHeikoS, your command worked17:00
OXPHOSlambday: hey any idea what may go wrong with global variable in benchmark?17:01
@HeikoSsanuj: great!17:01
@HeikoSwiking: around?17:03
-!- c4goldsw [c1a99ae1@gateway/web/cgi-irc/kiwiirc.com/ip.193.169.154.225] has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client]17:03
Saurabh7hello17:06
Saurabh7HeikoS: hi17:06
@HeikoSSaurabh7: hey17:07
@HeikoShow are things?17:07
@HeikoSshall we chat a bit in a moment?17:07
@HeikoSjust getting a coffee :)17:07
@HeikoSSaurabh7: merged the xvalidation btw17:07
Saurabh7ok17:07
Saurabh7great!17:07
sanujHeikoS, will you be around after 45 mins?17:08
Saurabh7HeikoS: ahvent treid that eigen method ou mentioned17:08
@HeikoSsanuj: yes17:08
sanuji need to have dinner17:08
@HeikoSsanuj: sure, enjoy!17:08
@HeikoSSaurabh7: yeah that would be good17:08
Saurabh7but we dont use the whole matrix17:08
Saurabh7X17:08
@HeikoSas it really simplifies the code17:08
Saurabh7only a subset of it17:08
@HeikoSmmh17:08
@HeikoSlet me see17:08
@HeikoShttps://github.com/jbowlan/pyl1ls/blob/master/lars.py#L10417:08
@HeikoSso reading this, they definitely use off the shelf updates17:09
@HeikoSwhere you have a Cholsyky matrix, say R17:09
@HeikoSand then you add x*x.T to the original matrix and want the new R17:09
@HeikoSSaurabh7: the other thing is17:10
@HeikoSSaurabh7: maybe we should not copy the matrix around all the time17:10
Saurabh7HeikoS: yes but  in shougn right now its a bit different17:10
@HeikoSI see17:10
@HeikoSSaurabh7: ok, I am still reading things17:10
Saurabh7its onl checks columns which are in m_active_17:10
Saurabh7so right now that choolesky_insert is hard coded for it17:11
@HeikoSokok17:11
@HeikoScool then17:11
@HeikoSSaurabh7: and what are you currently working on there?17:12
Saurabh7but in reality it should be possible ?17:12
@HeikoSno it is ok for now17:12
@HeikoSlets look at benchmark17:12
@HeikoSand then decide17:12
Saurabh7HeikoS: almost done only 2 more solves to be replaced17:12
@HeikoSis there an implementation that uses a similar solver?17:12
@HeikoSSaurabh7: cool17:12
@HeikoSSaurabh7: is the code guarded by unit test?17:12
Saurabh7i added more17:12
@HeikoSthat is, are you making sure you are not introducing bugs?17:12
Saurabh7for the cholesky insert17:12
@HeikoSI think there were some tests before right?17:13
@HeikoSfor the full algo17:13
Saurabh7so thats definetly good17:13
Saurabh7HeikoS:  yes there were17:13
@HeikoSokok17:13
@HeikoSso as long as they still pass all good17:13
@HeikoSand benchmark?17:13
Saurabh7HeikoS: I ran benchamrk framwework17:13
Saurabh7its was bugged i send a fix17:13
@HeikoS(havent read all email yet btw)17:13
@HeikoSwhat was bugged?17:14
Saurabh7the way input was handled17:14
Saurabh7HeikoS: anyways the runtimes are very low even for big dataset17:14
Saurabh7I have values I wil psot there17:15
shogun-buildbotbuild #2895 of bsd1 - libshogun is complete: Failure [failed configure]  Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/2895  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Saurabh7 <saurabh.mahindre@gmail.com>17:15
@HeikoSSaurabh7: what did you compare against?17:16
@HeikoSyeah post17:16
Saurabh7scikit17:16
@HeikoSSaurabh7: cool very curious about the times, give them :)17:16
@HeikoSI will grab a coffee in the meantime17:16
Saurabh7okie17:16
shogun-buildbotbuild #33 of xenial - libshogun is complete: Failure [failed compile]  Build details are at http://buildbot.shogun-toolbox.org/builders/xenial%20-%20libshogun/builds/33  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Saurabh7 <saurabh.mahindre@gmail.com>17:17
shogun-buildbotbuild #23 of FC23 - libshogun - aarch64 is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/FC23%20-%20libshogun%20-%20aarch64/builds/23  blamelist: Esben Sorig <esben@sorig.eu>, Heiko Strathmann <heiko.strathmann@gmail.com>17:19
@lambdayOXPHOS: what is going wrong with the global variable?17:39
@HeikoSSaurabh7: back17:39
@HeikoSso what about the times?17:39
@HeikoSgot them?17:39
@lambdayOXPHOS: it worked with the unit-tests, right?17:39
Saurabh7HeikoS: posted some17:40
@HeikoSah lemme check17:40
@HeikoSwhich link is it?17:40
Saurabh7need to add scikit for one dataset too17:40
Saurabh7https://github.com/shogun-toolbox/shogun/pull/324317:40
OXPHOSlambday: yes!17:43
sanujHeikoS, can you merge https://github.com/shogun-toolbox/shogun-data/pull/9617:43
OXPHOSlambday: if I do sg_linalg.set_cpu_backend(..), it gives segementation fault17:44
@lambdayOXPHOS: did you call init_shogun() ?17:44
@HeikoSsanuj:  done!17:44
sanujthanks17:44
OXPHOSlambday: ...nooooooo17:44
@lambdayOXPHOS: there you go :)17:45
@lambdayOXPHOS: do one thing. In the constructor of the struct Data, call init_shogun_with_defaults().. and in the destructor, call exit_shogun() :)17:45
OXPHOSlambday: sure. thanks!17:45
@lambdayOXPHOS: nw :)17:46
@HeikoSSaurabh7: these look good17:46
@HeikoSare there also some for which it takes longer than 10s17:46
@HeikoSSaurabh7: another question: does our LARS allow to extract the regularisation path easily?17:46
@HeikoSSaurabh7: and what are the next steps?17:47
Saurabh7HeikoS: mayb i can find one for which bigger runtimes, but htey migh tnot be meaningful dataset17:48
Saurabh7HeikoS: yes i think so17:48
@HeikoSno please use a meaningful one :)17:48
@HeikoSSaurabh7: you can also just generate data yourself17:48
@HeikoSwhich is LASSO friendly17:48
@wikingOXPHOS: so basically with the tags any you could store all the params right?17:48
@HeikoSthat is: a linear regression function with many dimensions17:48
@wikinglisitsyn: ^17:48
@HeikoSbut with only a few that contribute to the ys17:48
Saurabh7HeikoS: ok lets comapre on sythetic big one then17:49
OXPHOSwiking: yes i think so17:49
@HeikoSSaurabh7: do you see how I mean this?17:49
@HeikoSSaurabh7: maybe prepare a notebook that generates such data and share, then I can comment17:49
@wikingOXPHOS: ok so that'd be enough to have the serialization of that map17:49
sanujHeikoS, i forgot to do Math:init_random(1) ==> Math:init_random(0)17:49
sanuji hope that is fine17:49
Saurabh7HeikoS: only few variates should be included in final ?17:49
@HeikoSsanuj: dont worry aobut it17:50
@HeikoSSaurabh7: so the model is17:50
@HeikoSy=w*x17:50
@wikinglooooooooooooool17:50
* wiking reacts on the runtime of lars17:50
@HeikoSwiking: why? :)17:50
@wikingHeikoS: the diff17:51
@HeikoSSaurabh7: and x and w are high dimensional17:51
@HeikoSbut w has mostly zeros17:51
@wikingHeikoS: btw do we wanna do the cholesky from eigen, not linalg?17:51
@HeikoSSaurabh7: so you generate x from say a gaussian17:51
@HeikoSpick w to have only few non-zero components17:51
@HeikoSand then generate y17:51
@HeikoSwiking: its good right?17:51
@HeikoSwiking: but for these small problems, the difference is not too meaningful17:52
@HeikoSwiking: I am curious to see this for problems that take like a minute (or more) for shogun17:52
@HeikoSwiking: we want to use linalg17:52
@HeikoSbut as it is not ready17:52
@HeikoSwe can use eigen for now17:52
@wikingwhat do you mean not ready?17:52
@wiking i mean we could add to the current linalg no?17:52
@HeikoSwiking: yeah17:52
@HeikoSI think the factorisation is using linalg17:52
@wikingi mean the function call will not change17:52
@HeikoStrue17:53
@wikingjust how it is implemented in the backend17:53
@HeikoSSaurabh7: ^17:53
@wikingwhich is fine17:53
@HeikoSyeah I agree17:53
@wikingbut we dont wanna rewrite algos like lars17:53
@HeikoSSaurabh7: so we need a triangular solve in linalg17:53
@HeikoSwiking: doing this atm :)17:53
@HeikoSsince it was all hand-implemented17:53
@HeikoSall the dot products17:53
@HeikoSusing lapack and loops17:53
@wikingwe shoudl rather implement a cholesky as part of linalg17:53
@wikinghehehe17:53
@HeikoSwiking: it is there an dused17:54
@wikingi mean we could move those17:54
@HeikoSbut not the triangular solve17:54
@wikingto linalg :P17:54
@wikingbut i guess nobody wants to support that17:54
@HeikoSnop :)17:54
@HeikoSthere actually is some hybrid cholesky algorithms17:54
@HeikoSthat use GPU and CPU in a combined manner17:54
Saurabh7HeikoS: I am using eigen solve, jsut need to extract specific cols from X17:54
@HeikoSSaurabh7: you are using a triangular solve in eigen, no?17:55
Saurabh7yesyes17:55
@HeikoSso that should be in linalg17:55
Saurabh7ok linalg17:55
Saurabh7isee17:55
@HeikoSbut lets do it in a second step17:55
@HeikoSfirst finish all the rest17:55
@HeikoSas changing is easier then17:55
@HeikoSSaurabh7: any chance that multicore helps in the LARS implementation you recon?17:56
@HeikoSprobably not right?17:56
Saurabh7was thinking about it17:56
Saurabh7not anythign obvious tho17:56
Saurabh7maybe some of the dot products17:57
@HeikoSSaurabh7: so for the benchmark18:00
@HeikoScan you try this for multiple D, N, and #num_non_zero18:00
@HeikoSand post numbers for each?18:00
@HeikoSidentifying where shogun wins more and where less?18:00
@HeikoSand then I will tweet "shogun 10x faster LARS than sklearn with 10x less developers" :D18:01
@HeikoSwiking: ^18:01
Saurabh7HeikoS: sure :)18:02
Saurabh7gotta run for dinner now , might be back a bit later18:02
Saurabh7HeikoS: should be up by tmrw morn18:02
@HeikoSSaurabh7: cool, will catch up then18:03
shogun-buildbotbuild #24 of FC23 - libshogun - aarch64 is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/FC23%20-%20libshogun%20-%20aarch64/builds/24  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Saurabh7 <saurabh.mahindre@gmail.com>18:05
@wikingjust a sec18:07
@wiking:>>>>18:07
@wikingwe need to start autogen the benchmarks18:07
@HeikoSyep definitely18:08
@HeikoSto support such claims18:08
@HeikoSthat is also something nice to show to the amazon people18:08
@HeikoSsince they employ a few sklearn devs ;)18:08
@HeikoSwiking: any eta for that?18:08
@wikinghahaha18:09
lisitsynshogun has yandex devs!18:09
@wikinglisitsyn: devs?18:09
@wikingyou mean dev18:09
lisitsynDEV18:09
@wiking:D18:09
lisitsyn:D18:09
@HeikoSit is THE DEV18:09
@HeikoS(tm)18:09
@wikingi'm gonna look into it18:09
@wikingand start running benchmarks18:10
@wikingnow that we move off everything18:10
@HeikoSwiking: yeah the bot should be idling these days18:10
@wikingfrom the servers18:10
@wikingnot 100%18:10
@HeikoSthe aws seems to work quite smooth18:10
@wikingthe first tests are still on the boths18:10
@wikingHeikoS: had some hiccups18:10
@wikingbut yeah18:10
@HeikoSwiking: oh one more thing I wanted to ask about18:10
@wikingi'll move the devian tests as well18:10
@HeikoSrelease and gpl18:10
@wiking*debian18:10
@wikingyes18:10
@HeikoSI created a repo18:10
@wikingsaw it18:10
@HeikoSbut before I move things18:11
@HeikoSshould discuss briefly18:11
@HeikoShow to modify the build etc18:11
@HeikoSand dir structure18:11
@HeikoSetc18:11
@wikingwell18:11
@wikingi reckon we should just have an extra dir under src/shogun18:12
@wikingthat is a submodule18:12
@HeikoSyeah18:12
@HeikoSbut is it flat?18:12
@HeikoSor does it mirror the dir structure18:12
@wikingmmm doesn't matter18:12
@wikingcould be anything18:12
@HeikoSall files in there plain is a a bit ugly18:12
@wikingi guess it should be a bit hierarch18:12
@wikingbut maybe per project18:13
@wikingand not per type18:13
@wikingso not like18:13
@wiking'classifier'18:13
@wikingbut something like: 'implementation_x'18:13
@HeikoSI see18:13
@HeikoSyeah maybe18:13
@wikingbut htat's just an idea18:13
@HeikoSand we can just incrementally move thing18:14
@HeikoS1.) move file to gpl repo 2.) delete from main repo and chnage paths 3.) check build18:14
@HeikoS4.) repeat18:14
@HeikoSwiking: we will keep the switch in cmake I guess18:15
@wikingyes18:15
sanujHeikoS, i tried --> all_layers = all_layers.linear(50); all_layers = all_layers.softmax(2); all_layers = all_layers.done()18:19
sanujthe translation fails18:19
@HeikoSsanuj: whats the error?18:19
@HeikoSsanuj: does that work in Python?18:19
lisitsynlambday: hey18:20
@HeikoStranslation itself?18:20
@HeikoSsanuj: can you share the code and details, then I can check it locally18:20
lisitsynlambday: what's the matter with spaces :D18:20
lisitsynain't it super ugly to have18:20
lisitsynx=y18:20
sanujHeikoS, there is no error, the cpp file generated is incomplete because of problems in translation18:20
sanuji'll send you the code18:20
@HeikoSsanuj: I see18:20
@HeikoSyeah lemme check18:20
sanujit works in python18:20
@HeikoSsanuj: sanity check18:21
@HeikoSmeta examples dont use semicolons18:21
@HeikoSput every statement in nits own line18:21
sanujHeikoS, i think it is because these functions return a pointer to NeuralLayers18:21
sanujyeah, i didn't put semi colons18:22
@HeikoStranslation doesnt care about semantics18:22
@HeikoSso it cannot fail due to semantics18:22
sanujHeikoS, i had tried get_layer_parameters also18:22
sanujit returned SGVector<>*18:22
@HeikoSsanuj: send it18:22
sanujand failed18:22
@HeikoSsanuj:  that is not SWIG compativlew18:22
@HeikoScompatible18:22
sanujyeah, i'll send18:23
@HeikoSmaybe start with a python example and then translate that to .sg18:23
sanujokay18:23
sanujHeikoS, btw, when i run "make test" or "ctest", most of the python-modular tests fail http://pastebin.com/6RZdijXf18:24
sanujwe had discussed this earlier18:24
sanujmy PYTHONPATH = /home/sanuj/Projects/shogun_cb/buildpython/src/interfaces/python_modular18:24
sanujand LD_LIBRARY_PATH = /home/sanuj/Projects/shogun_cb/buildpython/src/shogun18:24
sanuji'm sure this is because of python318:27
sanuj"python distance_tanimoto_modular.py" but "python3 distance_tanimoto_modular.py18:27
sanuj" doesn't18:27
@HeikoSsanuj: what happens if you run any of the examples manually?18:27
@HeikoSwith what error?18:27
sanujHeikoS, ImportError: dynamic module does not define init function (PyInit__modshogun)18:28
OXPHOSHeikoS: I'm current doing in conf.py: 'cookbook_classifier_linearsvm': ('doc/cookbook/html/examples/classifier/linear_svm.html',      ''),18:28
OXPHOSHeikoS: too specific?18:28
@HeikoSsanuj: that is your python setup I am quite sure18:29
@HeikoSsanuj: not sure how to fix it18:29
@HeikoSdid you ask google?18:29
@HeikoSwiking might know as well ^18:29
@HeikoSOXPHOS: sorry, what? :18:29
@HeikoS:)18:29
OXPHOSHeikoS: u want a hyperlink from "see the linear SVM cookbook" right?18:30
OXPHOSthe only way I can think of is 'cookbook_classifier_linearsvm': ('doc/cookbook/html/examples/classifier/linear_svm.html',      ''),18:30
OXPHOSand in .sg :cookbook_classifier_linearsvm:`see the linear SVM cookbook`18:30
OXPHOS.rst18:31
@HeikoSyeah hyperlink would be good18:31
sanujwiking, ping pong18:31
@HeikoSOXPHOS: let me check, just a sec18:32
@HeikoSOXPHOS: whats the link again?18:32
sanujHeikoS, but why is "make test" using python318:32
@wikingpong pong18:32
OXPHOSHeikoS: https://github.com/shogun-toolbox/shogun/pull/3242  ?18:32
sanujwiking, on running python modular examples with python3 i get "ImportError: dynamic module does not define init function (PyInit__modshogun)"18:32
OXPHOSHeikoS I don't want to mislead sanuj since you mentioned..18:32
@wikingsanuj: what's the problem with python318:33
sanujwiking, most of the python-modular examples fail when i do "make test", i think it is because of python318:33
@wikingworks here http://buildbot.shogun-toolbox.org/builders/deb4%20-%20python3/builds/86218:33
@HeikoSOXPHOS: ah now I see18:34
sanujwhy am i getting that error18:34
@HeikoSOXPHOS: so if you "make cookbook" this, what does the link point to?18:34
@wikingsanuj: good question18:34
@HeikoSsanuj: I suspect it is you python setup, and that something is messed up between python2/318:35
@HeikoSyour shogun is not only compiled against one version or so18:35
sanujwiking, and why is "make test" using python3 and not python 2.718:35
OXPHOSHeikoS: actually is is "http://localhost:8000/doc/cookbook/html/examples/classifier/doc/cookbook/html/examples/classifier/linear_svm.htmlSee%20the%20linear%20SVM%20cookbook"18:36
@HeikoSsanuj: make test uses the python interpreter that swig finds18:36
@wikingsanuj: i need to see your cmake output18:36
@HeikoSOXPHOS: this is a bit more problematic, I just reliased, so the "latest" can stay for now18:36
@wiking|plz email18:36
@HeikoSOXPHOS: we need to change that later18:36
sanujHeikoS, my cmakecache.txt has PYTHON_PACKAGES_PATH             lib/python2.7/dist-packages18:36
@wikingas im about to sleep18:36
@HeikoSOXPHOS: so ignore my comment18:36
sanujwiking, okay :)18:36
OXPHOSHeikoS: okay..actually I thought it would be easy ;) thanks18:37
@HeikoSOXPHOS: we will need to have some notion of the current shogun version in there and that put that in the link18:38
@HeikoSbut we dont have that yet18:38
@HeikoSwill work on that after gsoc18:38
@HeikoSsince low priority18:38
@HeikoSOXPHOS: did i comment on anything else on the linear machine?18:38
@lambdaylisitsyn: I totally agree. I myself never use x=y :D but in Shogun we don't use that : https://github.com/shogun-toolbox/shogun/wiki/Code-style18:39
lisitsynlambday: but we don't really follow that18:39
OXPHOSHeikoS: yes. some phrases and the creation of the instance18:39
lisitsynsanuj: could you please ignore lambday :D :D :D18:39
@lambdaylisitsyn: then it shouldn't be in coding styles :D18:39
@lambdaylisitsyn: I try to follow that when pushing code to shogun anyway :D18:39
lisitsynlets change it!18:40
lisitsynx=y18:40
lisitsynwtf18:40
OXPHOSHeikoS: you mean we keep full address like http://. How about the rename stuff?  linear_svm --> see the linear SVM cookbook18:40
@lambdaylisitsyn: fine with me.18:41
@HeikoSx=y?18:41
@HeikoSwhat?18:41
@lambdayno spaces between operators and operands18:41
lisitsynHeikoS: lambday wants no spaces around =18:41
lisitsyn:)18:41
@HeikoSlisitsyn: thats what we have written down18:41
@HeikoSlisitsyn: should change it then18:41
@HeikoSOXPHOS: yeah keep full address to the "latest"18:41
@HeikoSrename works just like in html's href (I think)18:42
sanujlisitsyn, lambday rofl18:44
OXPHOSHeikoS: I think you didnt get me about kmeans..now the dataset and kmeans are both updated. but the kmeans integration dataset need to be merged first18:48
@HeikoSI see18:48
@HeikoSso I should merge?18:49
@HeikoSsorry too many emails ;)18:49
OXPHOSHeikoS: yep thanks18:49
OXPHOSHeikoS: I can tell that :)18:49
@HeikoSkmeans merged18:49
OXPHOSHeikoS: thanks!18:49
@HeikoSOXPHOS: often it helps if you just tell me what to do "HeikoS merge this [link]" ;)18:49
@HeikoSOXPHOS: your other data pr doesnt have any output18:50
@HeikoSneed to extract something18:50
OXPHOSHeikoS: sure;) I'll look into that. Yes I just generated some lables18:50
@HeikoScool18:51
@HeikoSanyone here knows whether JULIA supports SWIG these days?18:53
sanujHeikoS, gmm is failing https://travis-ci.org/shogun-toolbox/shogun/jobs/135628446#L263219:00
sanujfor ruby19:00
sanujcan't say about java and lua, the build terminated19:00
@HeikoSchecking19:06
@HeikoSah static call fun19:06
@HeikoSchecking19:06
@HeikoSsanuj:  I think ruby is easy to fix19:08
@HeikoSwill send a patch19:08
sanujHeikoS, okay thanks19:08
@HeikoSrestarting lua19:08
@HeikoScan you ping me on that?19:08
sanujHeikoS, i'm going to sleep19:09
@HeikoSsleep well19:09
sanujHeikoS, can you review my tags PR19:09
@HeikoSsanuj: I can, what to focus on?19:09
@HeikoSsanuj: is this for merging already?19:09
sanujit's ready to be merged apart from some comment by lambday19:09
@HeikoSokok19:09
sanujHeikoS, i'll look at it tomorrow19:09
sanujhttps://github.com/shogun-toolbox/shogun/pull/322119:10
@HeikoScool, see you!19:10
sanujcomment in the PR19:10
-!- sanuj [~sanuj@117.203.18.75] has quit [Ping timeout: 276 seconds]19:15
-!- GSqU [~GSqU@104.156.238.186] has joined #shogun19:39
-!- GSqU [~GSqU@104.156.238.186] has left #shogun []19:40
-!- lambday [8028b10a@gateway/web/freenode/ip.128.40.177.10] has quit []19:54
OXPHOSHeikoS: Hey, I didn't find anything to extract for multi-class logistic regression and multi-class linear machine19:55
OXPHOSvia the multi-class classifier directly19:56
-!- besser82 [~besser82@fedora/besser82] has quit [Ping timeout: 260 seconds]20:25
-!- besser82 [~besser82@fedora/besser82] has joined #shogun20:33
-!- mode/#shogun [+o besser82] by ChanServ20:33
-!- besser82 [~besser82@fedora/besser82] has quit [Ping timeout: 246 seconds]20:38
-!- besser82 [~besser82@fedora/besser82] has joined #shogun20:40
-!- mode/#shogun [+o besser82] by ChanServ20:40
@HeikoSrcurtin: jo20:43
@HeikoSOXPHOS: e.g. the labels20:43
@HeikoSOXPHOS: you can extract the binary labels for each class, or the multiclass label vector20:44
rcurtinHeikoS: hello there20:46
rcurtinin a meeting, might be slightly slow to respond20:46
@HeikoSrcurtin: hi man, how are things?20:46
rcurtinGSoC makes me busy, I always forget just how much work it is :)20:47
@HeikoSrcurtin: haha same here20:48
@HeikoSrcurtin: there is a PR for the benchmark thing on lars, just wanted to ask to merge as it fixes a bug20:48
rcurtinsure, is it ready to go?  you just made a comment on it pointing out that it might not be the right thing20:48
@HeikoSlemme check20:50
@HeikoSrcurtin: dont see your comment20:50
rcurtinI did not make a comment, you did :)20:51
rcurtin"Not sure I get this.  Definitely, it only makes sense to compare against [this one](http://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LassoLars.html) "20:51
@HeikoSah sorry20:52
@HeikoSSaurabh7:  around?20:54
@HeikoSrcurtin: no merge pls :)20:57
@HeikoSrcurtin: commented20:57
@HeikoSrcurtin: how is gsoc going for you guys?20:57
rcurtinHeikoS: it is going well, lots of code starting to come in now20:58
@HeikoSrcurtin: good code? ;)20:58
rcurtinyeah, it is great so far, but it takes a long time to check through it all and make sure it is good20:58
rcurtinI think the first few PRs are especially hard because you have to check style and all kinds of other little things before you trust the students fully and check their code in less detail20:59
rcurtinit looks like this channel is quite busy, I would guess you guys are having a similar experience :)21:00
@HeikoSrcurtin: yeah indeed21:00
@HeikoSrcurtin: yeah very busy21:00
@HeikoSwe are going for tighter mentoring this year21:00
@HeikoSas the projects are quite invovled21:00
@HeikoSnot just extensions21:00
@HeikoSrcurtin: you are working on a csv parser I see ?:)21:01
-!- travis-ci [~travis-ci@ec2-54-205-226-52.compute-1.amazonaws.com] has joined #shogun21:03
travis-ciit's lambday's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/13564701421:03
-!- travis-ci [~travis-ci@ec2-54-205-226-52.compute-1.amazonaws.com] has left #shogun []21:03
rcurtinHeikoS: yeah, tighter mentoring definitely pays off21:04
rcurtinafter some of our 2014 projects, it was like mid-2015 until I realized some fairly obvious issues with the usability of the code that was produced21:04
@HeikoSyeah21:04
@HeikoSsame21:04
@HeikoSand you also want people to stay around afterwards21:05
@HeikoSmore likely if they are integrated more21:05
rcurtinwe are indeed looking into writing a parser ourselves, because Armadillo's loading functionality doesn't really handle missing values and other things that are important in ML21:05
@HeikoSand more fun for everyone21:05
rcurtinI agree, definitely21:05
rcurtinboost::spirit seems really cool, but quite crazy21:05
rcurtinI don't like maintaining a parser ourselves, since it's just more code to take care off that can have bugs that will be our problem (instead of some other project's), but I am not sure we have a choice in this case21:05
@HeikoSrcurtin: you fine with boost as a dependency?21:06
@HeikoSwe're avoiding that here21:06
@HeikoSwith some pretty strong opinions on that21:07
@HeikoSbut our csv loader is buggy21:07
@HeikoS;)21:07
rcurtinHeikoS: yeah, I think the primary objection to boost is how huge it is, right?21:07
@HeikoSyes21:07
@HeikoScreates issues for portability21:07
@HeikoSit is a nightmare already for us21:07
rcurtinportability to what?21:07
@HeikoSkeeping all the builds green21:07
@HeikoSdifferent OS21:08
rcurtinwe have no issues with windows/os x/linux/freebsd21:08
@HeikoSwith different boost versions etc21:08
rcurtinanything embedded, probably you might have issues21:08
rcurtinah, yeah, we avoid that by being careful with the minimum accepted boost version and testing across boost versions21:08
rcurtinalthough, we don't really have a formal framework in place for that21:08
@HeikoShehe21:08
@HeikoSbtw have you seen our cookbooks?21:08
rcurtinI am *still* waiting for my company to mount our new jenkins server that they bought for us (I have been waiting since October)21:08
rcurtinI have glanced at some in the past, but not in depth21:08
@HeikoSwe are making every student write two a week atm21:08
@HeikoScheck shogun.ml/cookbook/latest21:08
rcurtinthese are like little tutorials?21:09
@HeikoSno21:09
@HeikoSmore like: API examples, integration test, and code snippets at once21:09
@HeikoSrcurtin: hey that is cool21:09
rcurtinah, yeah, that's right, they are also integration tests21:09
rcurtinthat is very nice, most people just want snippets to copy/paste and try21:09
@HeikoSyeah all executable code (in all languages)21:09
@HeikoSso travis runs all of them21:09
@HeikoSbut the pages only show a subset of the full listing21:09
@HeikoSwe recently got some amazon credits, and moved our buildbot there21:10
@HeikoSnow we have more space on our own servers for benchmarks21:10
rcurtindid they give them to you for free because you are an open source project?21:10
@HeikoSso will probably run your framework there21:10
rcurtinI seem to remember that they offer credits for stuff like that21:10
@HeikoSrcurtin: kind of, we applied21:10
@HeikoSand knew some people21:10
@HeikoSyeah21:10
@HeikoSpretty cool21:10
rcurtinknowing some people is always the best way :)21:10
@HeikoSwe are not nearly using all of them21:10
@HeikoSyep ;)21:11
@HeikoSthe old shogunners21:11
@HeikoSthey all did their phds together hehe21:11
@HeikoSI am way too young21:11
@HeikoSthe others even more21:11
rcurtinwhere did soeren end up?  is he still at tomtom, or did he go to amazon?21:11
rcurtinhe was the only old shogunner I ever met :)21:11
rcurtinis it shogunner or shoguneer? :)21:12
shogun-buildbotbuild #2896 of bsd1 - libshogun is complete: Failure [failed configure]  Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/2896  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>21:12
@HeikoStomtom21:12
@HeikoSthere is gunnar21:12
@HeikoShe supervised sören21:12
rcurtinah, okay, I did not realize he went to amazon21:13
shogun-buildbotbuild #34 of xenial - libshogun is complete: Failure [failed compile]  Build details are at http://buildbot.shogun-toolbox.org/builders/xenial%20-%20libshogun/builds/34  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>21:14
@HeikoSno he didnt21:14
rcurtinoh, okay, my comprehension level is low I guess :)21:14
@HeikoSbut both of them know the head of german amazon research21:14
@HeikoSyeah sorry I am too vague21:14
@HeikoSaws is really unreliable21:14
rcurtin:(21:15
@HeikoSI think viktor told you already21:15
@HeikoSbut it is ok for buildbotting21:15
rcurtindo you find the builds are randomly failing or something?21:15
@HeikoSI think they were21:15
@HeikoSand stalling21:16
rcurtinI remember we had a discussion about whether or not AWS was suitable for benchmarking, I think he disagreed with my opinion, but that is okay, my opinion is often wrong :)21:16
@HeikoSdisappearing21:16
rcurtinthat's really frustrating, but we have had the same problem with Jenkins sometimes21:16
@HeikoSI think he agrees with you21:16
@HeikoSnot suitable for benchmarking21:16
rcurtinI can't remember, too much has happened since then :)21:16
@HeikoSat that thats whatI understood21:16
@HeikoShehe21:16
rcurtinour whole build setup is stalled on Symantec getting this damn server mounted and publicly accessible21:16
@HeikoSI see21:16
@HeikoShard ?21:16
rcurtinbut it's just, there is so much paperwork to fill out, and the IT people are so slow, that we have been waiting for months21:17
-!- besser82 [~besser82@fedora/besser82] has quit [Ping timeout: 250 seconds]21:17
rcurtinthat's the thing, it shouldn't be difficult!  we just want two ports open (HTTP and SSH) but they have a giant list of requirements, the system has to be scanned, we have to submit forms for a firewall change request21:17
rcurtinbut they are not running the scan for some reason, and the request is not going through as a result21:17
rcurtinit's really stupid, too much process overhead at this company21:17
rcurtinfortunately I do not have to deal with the IT people directly, I have someone helping me, but even so, it is quite frustrating21:17
@HeikoSphew I can imagine this is nerve wrecking21:19
rcurtinit's irritating, because a lot of stuff can't proceed until that is done21:21
rcurtinI have enough to do in the mean time, but it's gone from being a monthly "I wonder when they will have that done" to a weekly "I wonder what the progress is" to a daily "why the hell is it not up yet"21:21
@HeikoShehe21:22
@HeikoSrcurtin: in that sense open source moves faster than companies21:22
@HeikoSat least if somebody who is motivated is around21:23
rcurtintruth21:23
@HeikoSrcurtin: well good luck with the rest of the first half of gsoc21:27
@HeikoSnice that the benchmarking is now a bit of a joint venture!21:27
rcurtinthanks, I am looking forward to what we come out with :)21:29
rcurtinboth for the first half of GSoC and the benchmarking system :)21:30
shogun-buildbotbuild #864 of deb4 - python3 is complete: Failure [failed install test python modular]  Build details are at http://buildbot.shogun-toolbox.org/builders/deb4%20-%20python3/builds/864  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>21:47
shogun-buildbotbuild #2897 of bsd1 - libshogun is complete: Failure [failed configure]  Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/2897  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>21:49
shogun-buildbotbuild #35 of xenial - libshogun is complete: Failure [failed compile]  Build details are at http://buildbot.shogun-toolbox.org/builders/xenial%20-%20libshogun/builds/35  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>21:52
-!- besser82 [~besser82@fedora/besser82] has joined #shogun21:53
-!- mode/#shogun [+o besser82] by ChanServ21:53
shogun-buildbotbuild #25 of FC23 - libshogun - aarch64 is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/FC23%20-%20libshogun%20-%20aarch64/builds/25  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>21:53
@HeikoSshogun-buildbot: force build 'deb4 - python3'21:54
@HeikoSforce build --branch=develop 'deb4 - python3'21:54
@HeikoSshogun-buildbot:  force build --branch=develop 'deb4 - python3'21:54
shogun-buildbotThe build has been queued, I'll give a shout when it starts21:54
-!- HeikoS [~heiko@nat-182-210.internal.eduroam.ucl.ac.uk] has quit [Quit: Leaving.]21:56
-!- lambday [6d9941c0@gateway/web/freenode/ip.109.153.65.192] has joined #shogun21:59
-!- mode/#shogun [+o lambday] by ChanServ21:59
@lambdayOXPHOS: hola22:00
@lambdayOXPHOS: is the global sg_linalg working now?22:00
@lambdayin the benchmark?22:00
@lambdayOXPHOS: also, did you get OpenCL to work locally?22:00
OXPHOSlambday: should be..Im about to run it22:01
OXPHOSlambday: it works, just didn't work with -l22:01
@lambdayOXPHOS: can you check your viennacl cmake output? it should be able to detect whether OpenCL is working or not22:01
@lambdayeven Shogun's cmake should be able to do that22:02
OXPHOSIs it in CMakeOutput.log?22:04
OXPHOSlambday ^22:04
OXPHOSapparently it didn't recognize init_shogun_with_defaults();22:05
OXPHOSlemme see22:05
OXPHOSworks now. #include22:07
@lambdayOXPHOS: yeah.. init.h22:07
@lambdayOXPHOS: also, about the benchmark, me and Heiko discussed a bit today..22:08
@lambdayOXPHOS: so basically what would be interesting to see is that, how much we pay for the flexibility22:08
@lambdayOXPHOS: please check if you can see this  -http://cosketch.com/Rooms/kymwluv22:08
OXPHOSyes lol22:09
@lambdaywe expect a plot like this22:09
-!- sonne|osx [~sonne@x4db3bf36.dyn.telefonica.de] has joined #shogun22:09
@lambdaypardon my drawing skills :D22:09
OXPHOSit's cute :)22:09
@lambdayhaha22:09
@lambdayso notice the diagram.. the overhead for the flexibility should go down as we increase the dimension of the vector22:10
@lambdayso, in practice, as long as the vector has a sufficiently large dimension, we don't pay anything extra..22:11
@lambdaywe want to know what that dimension is22:11
@lambdaysee the point?22:11
@lambdayhaha doodling :D22:11
lisitsynhow's my drawing22:11
@lambdaybetter than mine :D22:11
@lambdaythat was supposed to be a ;)22:12
OXPHOSlambday: a. SGVector -map to Eigen Vector - call Eigen dot ?22:12
@lambdayOXPHOS: yeah..22:13
OXPHOSbut this is what we're doing in CPUBackend.dot now..22:13
@lambdayso you should include the creation of map in the benchmark as well..22:13
@lambdayOXPHOS: of course.. that's the point22:13
lisitsynthe pigder!22:14
lisitsynspipig!22:14
@lambdaylisitsyn: that looks like captain kirk22:14
@lambdaypiggified version22:14
@lambdaylisitsyn: hahaha..22:15
OXPHOShaha22:15
lisitsynDONE22:16
@lambdaylisitsyn: I did something too xD22:17
@lambdaylisitsyn: hahaha22:17
@lambdayOXPHOS: so, do you copy? over.22:19
OXPHOSlambday: roger that22:19
shogun-buildbotbuild forced [ETA 5m47s]22:20
shogun-buildbotI'll give a shout when the build finishes22:20
@lambdayshogun-buildbot: don't shout, just dance22:20
@lambdayshogun-buildbot: dance22:20
shogun-buildbot<(^.^<)22:20
shogun-buildbot<(^.^)>22:20
shogun-buildbot(>^.^)>22:20
shogun-buildbot(7^.^)722:20
shogun-buildbot(>^.^<)22:20
@lambdaylisitsyn: so about the SGObject.. the subclasses register params, right?22:21
@lambdayand then the serialization works with the hashmap from the base class?22:21
shogun-buildbotbuild #865 of deb4 - python3 is complete: Success [build successful]  Build details are at http://buildbot.shogun-toolbox.org/builders/deb4%20-%20python3/builds/86522:25
shogun-buildbotbuild #2898 of bsd1 - libshogun is complete: Failure [failed configure]  Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/2898  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>22:26
shogun-buildbotbuild #36 of xenial - libshogun is complete: Failure [failed compile]  Build details are at http://buildbot.shogun-toolbox.org/builders/xenial%20-%20libshogun/builds/36  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>22:28
-!- lambday [6d9941c0@gateway/web/freenode/ip.109.153.65.192] has quit [Ping timeout: 250 seconds]22:30
shogun-buildbotbuild #26 of FC23 - libshogun - aarch64 is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/FC23%20-%20libshogun%20-%20aarch64/builds/26  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>22:34
lisitsynlambday: no need to register anything, they are directly in the map22:40
lisitsynsubclasses could set up default values22:40
-!- sonne|osx [~sonne@x4db3bf36.dyn.telefonica.de] has quit [Quit: sonne|osx]23:09
shogun-buildbotbuild #27 of FC23 - libshogun - aarch64 is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/FC23%20-%20libshogun%20-%20aarch64/builds/27  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>23:13
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun23:17
-!- mode/#shogun [+o HeikoS] by ChanServ23:17
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has quit [Client Quit]23:18
-!- besser82 [~besser82@fedora/besser82] has quit [Ping timeout: 246 seconds]23:40
-!- lambday [6d9941c0@gateway/web/freenode/ip.109.153.65.192] has joined #shogun23:44
-!- mode/#shogun [+o lambday] by ChanServ23:44
--- Log closed Tue Jun 07 00:00:29 2016
« Sunday, 2016-06-05 Index Tuesday, 2016-06-07 »

Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!