Open in new window / Try shogun cloud
--- Log opened Sun Jun 30 00:00:10 2013
-!- zxtx [~zv@] has joined #shogun00:18
-!- zxtx [~zv@] has quit [Ping timeout: 252 seconds]00:35
-!- zxtx [~zv@] has joined #shogun00:46
-!- zxtx [~zv@] has quit [Ping timeout: 252 seconds]00:53
-!- iglesiasg [] has joined #shogun01:33
-!- mode/#shogun [+o iglesiasg] by ChanServ01:33
-!- iglesiasg [] has quit [Quit: leaving]02:14
-!- foulwall` [~user@2001:da8:215:6100:e9db:38fb:2188:6960] has joined #shogun02:48
-!- pickle27 [] has quit [Quit: Leaving]03:07
shogun-buildbotbuild #443 of nightly_default is complete: Failure [failed test]  Build details are at
-!- zxtx [] has joined #shogun04:41
-!- foulwall` [~user@2001:da8:215:6100:e9db:38fb:2188:6960] has quit [Ping timeout: 264 seconds]05:40
-!- foulwall` [~user@2001:da8:215:6100:54f2:2753:812b:2f25] has joined #shogun06:41
-!- nube [~rho@] has joined #shogun07:52
-!- lambday [67157d4c@gateway/web/freenode/ip.] has joined #shogun08:54
lambdaysonney2k: moin moin :)08:54
-!- foulwall` [~user@2001:da8:215:6100:54f2:2753:812b:2f25] has quit [Ping timeout: 264 seconds]09:06
-!- nube [~rho@] has quit [Quit: Leaving.]09:11
@sonney2klambday, hehe moin moin :D09:49
lambdaysonney2k: how are you? :)09:49
@sonney2klambday, busy as usual09:50
@sonney2kbut happy that workshop organization is going well :D09:50
lambdaysonney2k: when is it?09:50
-!- foulwall` [~user@2001:da8:215:6100:5093:2b3e:1eeb:ea7b] has joined #shogun09:52
@sonney2klambday, 12-1409:54
@sonney2kfoulwall`, you have to adjust the gp example a bit the syntax changed09:55
lambdaysonney2k: looking forward to the video recording.. :)09:56
-!- nube [~rho@] has joined #shogun10:25
-!- van51 [] has joined #shogun10:33
-!- iglesiasg [] has joined #shogun11:23
-!- mode/#shogun [+o iglesiasg] by ChanServ11:23
-!- foulwall` [~user@2001:da8:215:6100:5093:2b3e:1eeb:ea7b] has quit [Ping timeout: 264 seconds]11:38
@sonney2klambday, I hope we can get some equipment for that12:12
@sonney2kvan51, did what I wrote make sense?12:16
van51sonney2k: yeap12:17
van51sonney2k: what I didn't quite get is for the nnz_features, why would the string length be better12:17
van51sonney2k: what I was/am thinking of doing there is to use the converter to transform the document to a SGSparseVector and store it and iterate it12:21
@sonney2kvan51, no the point is to do that on-the-fly12:25
@sonney2kvan51, and the nnz_features was IIRC just an estimation so an upper bound is fine12:25
van51sonney2k: but you can't iterate the features on the fly12:25
van51the nnz* features on the fly12:25
@sonney2kwhy not?12:25
van51you may get the same one again and again12:25
@sonney2kthat is ok12:26
van51and not in the correct order12:26
@sonney2kalso ok12:26
van51ok then :P12:26
van51these were my arguments12:26
@sonney2kjust think of it12:26
@sonney2kwill then just add the thing multiple times12:26
van51sonney2k: ah wait12:26
@sonney2kand dense_dot same thing12:26
van51sonney2k: I'm not talking about these methods12:26
@sonney2kignore the feature iterators12:26
van51sonney2k: but for the iterators12:27
@sonney2kjust do SG_NOT_IMPLEMENTED12:27
van51sonney2k: ok12:27
van51sonney2k: although we could iterate just the sparsevector there12:28
van51sonney2k: but I remember that the iterator is seldom used12:28
van51sonney2k: also, how important is the dot() method?12:29
van51sonney2k: should I also remove the converter from there?12:29
@sonney2kvan51, sufficient as is12:30
van51sonney2k: ok then12:30
van51sonney2k: I'll update the PR soon and start working on a unit test after that12:30
@sonney2kvan51, the iterator was intended for L1 methods12:31
@sonney2kvan51, yeah make a toy example give reasonable results and compare to the converter12:31
van51sonney2k: ok! I'll let you know how it goes12:32
-!- iglesiasg [] has quit [Quit: leaving]12:42
van51sonney2k: I updated the PR12:50
van51sonney2k: I realize that reviewing so many PRs and stuff may be a bit too much work for you12:53
van51sonney2k: should I ask Olivier and Benoit to start reviewing my PRs, as well?12:53
van51sonney2k: and Quoc later on12:53
-!- foulwall` [~user@2001:da8:215:6100:e193:4467:a5b4:a60] has joined #shogun12:58
foulwall`ping sonney2k, ping lisitsyn14:15
-!- foulwall` [~user@2001:da8:215:6100:e193:4467:a5b4:a60] has left #shogun ["ERC Version 5.3 (IRC client for Emacs)"]14:19
-!- van51 [] has quit [Quit: Leaving.]14:47
-!- nube [~rho@] has quit [Quit: Leaving.]15:38
-!- nube [~rho@] has joined #shogun15:43
-!- van51 [] has joined #shogun16:29
-!- nube [~rho@] has quit [Quit: Leaving.]16:35
-!- nube [~rho@] has joined #shogun17:09
-!- lambday [67157d4c@gateway/web/freenode/ip.] has quit []17:39
-!- pickle27 [] has joined #shogun18:09
-!- nube [~rho@] has quit [Quit: Leaving.]18:16
-!- foulwall [~user@2001:da8:215:c252:914b:a23f:a39f:5522] has joined #shogun18:31
-!- foulwall [~user@2001:da8:215:c252:914b:a23f:a39f:5522] has quit [Quit: sleep]19:19
-!- gsomix [~gsomix@] has joined #shogun20:03
gsomixgood evening20:04
gsomixsonney2k, lisitsyn finally I'm with internet and can start work.20:04
lisitsyngsomix: good20:04
@sonney2kgsomix, hey there!20:19
gsomixsonney2k, hey.20:19
gsomixsonney2k, I'm here now 55.923552,37.52496220:21
@sonney2kgsomix, a village!20:27
@sonney2k... with internet!20:27
@sonney2kvan51, alright what is the status?20:28
gsomixsonney2k, not village, science city with MIPT.
-!- iglesiasg [] has joined #shogun20:33
-!- mode/#shogun [+o iglesiasg] by ChanServ20:33
van51sonney2k: it's going pretty good I think20:34
van51sonney2k: if you don't find something wrong with that's pushed already I just have to finish a unit test case20:35
van51sonney2k: and I'll get started on an example20:35
@sonney2kvan51, feedback send20:43
@sonney2kfix it and I can merge20:43
@sonney2kgsomix, alright so now you have time right?20:44
van51sonney2k: cool, thank you20:44
@sonney2kgsomix, if so then please finish of your first PR.. add tests! and use van51's *Tokenizer classes!20:44
gsomixsonney2k, yep. now  getting familiar with gmock and unit-tests.20:46
gsomixsonney2k, Tokenizer for what?20:47
@sonney2kgsomix, string splitting20:48
gsomixsonney2k, for LineReader and CircularBuffer? or for further readers?20:48
@sonney2kgsomix, for the ascii based readers it makes sense. also for .csv splitting etc20:52
gsomixsonney2k, ah, ok. it might be very useful. I will look into the van51's code.20:53
@sonney2kgsomix, alright so do the tests & get your PR ready then lets talk again :)21:00
gsomixsonney2k, ok21:01
-!- shogun-notifier- [] has joined #shogun21:12
shogun-notifier-shogun-demo: foulwall :master * b55450e / / (20 files):
shogun-notifier-shogun-demo: 1. more clear js,did a lot refractoring. 2. abstract get_kernel method21:12
shogun-notifier-shogun-demo: out for reuse. 3. add a full support for heatmap with contour. 4.21:12
shogun-notifier-shogun-demo: abstract the json interactive out, for reuse. 5. merge dvalcarce's21:12
shogun-notifier-shogun-demo: binary classify demo in. 6. add a mechanism for auto button code21:12
shogun-notifier-shogun-demo: generation. 7. fixed a bug which in clustering demo, when user click21:12
shogun-notifier-shogun-demo: Soeren Sonnenburg :master * bf82dc7 / / (20 files):
shogun-notifier-shogun-demo: Merge pull request #9 from foulwall/refractor21:12
shogun-notifier-shogun-demo: a lot of refractor.21:12
shogun-notifier-shogun-demo: foulwall :master * 8fce12d / / (5 files):
shogun-notifier-shogun-demo: 1. merged the perceptron demo. since in enhanced framework, when21:13
shogun-notifier-shogun-demo: creating this demo, no boring js or html needed, just write a python21:13
shogun-notifier-shogun-demo: script to tell how to do calculation. 2. make an individual fetch_data21:13
shogun-notifier-shogun-demo: package, for read typical data from http request.21:13
shogun-notifier-shogun-demo: Soeren Sonnenburg :master * 8881aac / / (5 files):
shogun-notifier-shogun-demo: Merge pull request #10 from foulwall/perceptron21:13
shogun-notifier-shogun-demo: Perceptron21:13
-!- van51 [] has quit [Quit: Leaving.]21:44
-!- van51 [] has joined #shogun21:49
@iglesiasgI hope you guys will be supporting Spain tonight :)22:05
lisitsyniglesiasg: haven't heard they are playing tonight :D22:05
@iglesiasgin two hours!22:05
-!- gsomix [~gsomix@] has quit [Quit: Leaving]22:10
pickle27hey guys22:30
pickle27I've got a weird bug that I haven't been able to get for a while22:30
pickle27I have an eigen map that isn't working and is causing a malloc memory corruption22:31
pickle27its wierd because I create an SGMatrix create a map on it, perform some math, all of which is fine22:31
pickle27then I pass the SGMatrix to a function and its the eigen map inside the func which fails22:32
pickle27the SGMatrix is fine though I just printed it22:32
pickle27iglesiasg: lisitsyn ^22:32
pickle27any ideas?22:33
lisitsynpickle27: hey22:33
lisitsyncan't say anything specific22:33
lisitsynlets see code ;)22:33
lisitsynwell actually22:34
lisitsynvalgrind detects such things - may be we can check its output too22:34
pickle27here is the code22:35
pickle27so when using valgrind it works ..22:37
pickle27something wierd is up because when I save the matrix to file it also worked22:37
lisitsynpickle27: may be remove references here22:38
lisitsynSGMatrix<float64_t> tdsep(SGMatrix<float64_t> &x, SGVector<float64_t> &sel)22:38
lisitsynyou don't get real advantage here but it may be unsafe22:39
lisitsynit shouldn't though22:39
pickle27I think I only added then to try and fix this22:39
pickle27with valgrind my the map prints but then the program stalls22:42
pickle27yeah this is weird, now that I can run this code (using valrind) I was able to debug the rest of it and now valgrind says no errors22:49
pickle27well nvm it says no leaks but there looks like some errors22:50
lisitsynpickle27: I see no reason to fail we need to force it to fail somehow22:50
lisitsynpickle27: it is not reference thing I am sure22:51
lisitsynpickle27: try to reproduce it under valgrind, its output could tell a lot22:52
pickle27found one thing22:53
pickle27lisitsyn: my SGVector was too small, silly mistake22:53
lisitsynohh yes22:53
lisitsyn4 vs 322:53
pickle27lisitsyn: that might have been it22:54
pickle27lisitsyn: strange how it manifested22:54
pickle27lisitsyn: cause yeah it runs now22:54
pickle27thanks! I never think to use valgrind but its pretty handy22:57
lisitsynpickle27: always use it when meet segfault  ;)22:58
-!- iglesiasg [] has quit [Quit: leaving]23:01
-!- van51 [] has left #shogun ["PING 1372627855"]23:30
shogun-notifier-shogun: Roman Votyakov :develop * a59edcb / src/shogun/ (24 files):
shogun-notifier-shogun: update documentation of gp framework23:37
shogun-notifier-shogun: Sergey Lisitsyn :develop * 615ef2d / src/shogun/ (24 files):
shogun-notifier-shogun: Merge pull request #1199 from votjakovr/feature/gp_refactoring23:37
shogun-notifier-shogun: update documentation of gp framework23:37
shogun-buildbotbuild #1163 of bsd1 - libshogun is complete: Failure [failed test_1]  Build details are at  blamelist: Roman Votyakov <>23:55
--- Log closed Mon Jul 01 00:00:12 2013