--- Log opened Fri Mar 07 00:00:11 2014 | ||
--- Day changed Fri Mar 07 2014 | ||
@HeikoS | dhruv13: what are the most common use cases? | 00:00 |
---|---|---|
dhruv13 | HeikoS: KNN Search, afaik | 00:00 |
@HeikoS | dhruv13: thats also the purpose of this right? | 00:00 |
@HeikoS | could you send me the link of the task? | 00:00 |
dhruv13 | HeikoS: https://github.com/shogun-toolbox/shogun/issues/1936 | 00:01 |
@HeikoS | dhruv13: thanks! I lost a bit track of all those tasks :)( | 00:01 |
dhruv13 | HeikoS: what's troubling me is that i'm unfamiliar with how data is passed around to learning methods | 00:02 |
@HeikoS | dhruv13: ah I see | 00:02 |
@HeikoS | dhruv13: so you should work against CFeatures objects | 00:02 |
@HeikoS | if possible | 00:02 |
dhruv13 | HeikoS: for a rough version, can i just assume numeric data? | 00:02 |
@HeikoS | dhruv13: would be better to work in general against the CFeatures interface, but obviously some algorithms can only work in R^N | 00:03 |
@HeikoS | dhruv13: maybe talk to iglesias about this | 00:03 |
dhruv13 | HeikoS: KNN should not though right? | 00:03 |
dhruv13 | yeah | 00:03 |
@HeikoS | dhruv13: what do you mean? | 00:03 |
dhruv13 | KNN can be applied to any string or vector feature type data | 00:04 |
@HeikoS | dhruv13: CDistance is also implemented against generic CFeatures | 00:04 |
@HeikoS | dhruv13: yeah there are distances on strings etc | 00:04 |
@HeikoS | dhruv13: and also on sparse data | 00:04 |
@HeikoS | dhruv13: even though it doesnt make sense then :) | 00:04 |
dhruv13 | the current cover tree implementation supports only EuclideanDistance, instead of the generic distance | 00:05 |
dhruv13 | which is what is confusing me | 00:05 |
@HeikoS | dhruv13: KNN works against general features also | 00:05 |
@HeikoS | dhruv13: ah I see let me check | 00:05 |
@HeikoS | dhruv13: which class and file? | 00:05 |
dhruv13 | /src/shogun/lib/{JLCoverTreePoint.h}{JLCoverTree.h}; | 00:07 |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 00:07 | |
shogun-notifier- | shogun: pl8787 :develop * 2973470 / tests/unit/classifier/svm/LibLinear_unittest.cc: https://github.com/shogun-toolbox/shogun/commit/2973470a919c5a115877ac2727fe83480ee86675 | 00:07 |
shogun-notifier- | shogun: For #1935 LibLinear unit test added. | 00:07 |
shogun-notifier- | shogun: pl8787 :develop * c54553f / src/shogun/classifier/svm/LibLinear.cpp: https://github.com/shogun-toolbox/shogun/commit/c54553fdd05299f475ef6fbd55ebc14206475f33 | 00:07 |
shogun-notifier- | shogun: Fix the function solve_l2r_lr_dual failed, when set bias is true. | 00:07 |
shogun-notifier- | shogun: pl8787 :develop * 51fdd04 / src/shogun/classifier/svm/LibLinear.cpp: https://github.com/shogun-toolbox/shogun/commit/51fdd0417193132b3533a8e804c6214f8e476b3c | 00:07 |
shogun-notifier- | shogun: Seting prob->n correctly. | 00:07 |
shogun-notifier- | shogun: pl8787 :develop * a51badb / tests/unit/classifier/svm/LibLinear_unittest.cc: https://github.com/shogun-toolbox/shogun/commit/a51badbdcd8910e9b314fafeaa62946780638031 | 00:07 |
shogun-notifier- | shogun: Add 7 Tests. | 00:07 |
shogun-notifier- | shogun: pl8787 :develop * 35cfec9 / tests/unit/classifier/svm/LibLinear_unittest.cc: https://github.com/shogun-toolbox/shogun/commit/35cfec9db5c541b59db2f308f730d041cc64af97 | 00:07 |
shogun-notifier- | shogun: Merge the common code into a function. | 00:07 |
shogun-notifier- | shogun: pl8787 :develop * 840e7f1 / tests/unit/regression/lars_unittest.cc: https://github.com/shogun-toolbox/shogun/commit/840e7f1d57d3519fb100b342a75f0e6551f43658 | 00:07 |
shogun-notifier- | shogun: Add 2 LAR test case. | 00:07 |
shogun-notifier- | shogun: pl8787 :develop * 9ca6189 / tests/unit/regression/lars_unittest.cc: https://github.com/shogun-toolbox/shogun/commit/9ca6189f57a0f89ad849a5ed962b34439eeb5229 | 00:07 |
shogun-notifier- | shogun: Add 2 LAR test case. | 00:07 |
@HeikoS | dhruv13: I dont get it, it looks generic to me, which lines are you refering to? | 00:08 |
dhruv13 | one moment | 00:09 |
dhruv13 | HeikoS: http://www.shogun-toolbox.org/doc/en/3.0.0/classshogun_1_1CDistance.html#ac40d441939fab88fc6d96159f7b535c0 | 00:10 |
dhruv13 | That's the documentation for CDistance | 00:10 |
dhruv13 | and i'm referring to line 152-153 in JLCoverTreePoint.h | 00:10 |
@HeikoS | dhruv13: I see | 00:11 |
dhruv13 | the distance_upper_bounded function | 00:11 |
@HeikoS | dhruv13: well the interface is general | 00:11 |
@HeikoS | its just that this function is only efficient for EuclideanFeatures | 00:11 |
@HeikoS | and inefficient otherwise, since it just returns the distance | 00:11 |
@HeikoS | so keep your implementation in the same way | 00:12 |
@HeikoS | if it only works for Euclidean for now thats fine, but pls keep the interface general | 00:12 |
dhruv13 | okay | 00:12 |
@HeikoS | dhruv13: so currently if we implement this upper bound thing for another distance, the cover tree would immediately work for that too | 00:12 |
@HeikoS | dhruv13: see what I mean? | 00:12 |
dhruv13 | HeikoS: yeah... | 00:13 |
@HeikoS | dhruv13: but pls synch and discuss this with fernando, he knows more than I do :) | 00:13 |
dhruv13 | understood! | 00:13 |
-!- thoralf [~thoralf@p4FEC8B68.dip0.t-ipconnect.de] has joined #shogun | 00:13 | |
dhruv13 | okay! thanks! | 00:13 |
thoralf | Hey. | 00:13 |
shogun-buildbot | build #2746 of deb1 - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun/builds/2746 blamelist: pl8787 <pl8787@gmail.com> | 00:14 |
dhruv13 | HeikoS: just to clarify, i can use an intermediate representation for data as long as i implement the necessary conversion code? | 00:14 |
dhruv13 | like JLCoverTreePoint does, with v_array? | 00:14 |
@HeikoS | thoralf: hi! | 00:15 |
@HeikoS | dhruv13: could you be more precise, I dont get what you want to do | 00:15 |
@HeikoS | dhruv13: whats your intermediate representation? | 00:15 |
dhruv13 | HeikoS: I need to sort the data points on the basis of each feature | 00:16 |
shogun-buildbot | build #2747 of deb1 - libshogun is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun/builds/2747 | 00:16 |
dhruv13 | HeikoS: so i'll create an array of data points, and move pointers around to sort it as i need to | 00:16 |
dhruv13 | based on the axis i'm considering | 00:16 |
dhruv13 | as far as i know, none of the data structures in shogun let me do this right now | 00:17 |
dhruv13 | i will write a constructor for this array to accept a CFeature object | 00:18 |
@HeikoS | dhruv13: what about adding such methods to CFeatures or CDistance? | 00:18 |
@HeikoS | dhruv13: so that the implementation works against a generic interface? | 00:18 |
@HeikoS | dhruv13: and then you implement the methods you need for dense/sparse/string? | 00:18 |
dhruv13 | HeikoS: yeah, i was actually thinking the same thing | 00:18 |
@HeikoS | dhruv13: thats best in terms of sustainability | 00:19 |
@HeikoS | dhruv13: all those local hacks usually bite use a few years later ;) | 00:19 |
dhruv13 | HeikoS: ha ha | 00:19 |
@HeikoS | dhruv13: so those things are also interesting from a software engineering perspective | 00:19 |
dhruv13 | HeikoS: for a rough patch, can i submit the outer data structure? Once i get the kd tree to work properly, I will modify CFeature accordingly... | 00:21 |
thoralf | HeikoS: Moin! | 00:21 |
@HeikoS | dhruv13: you can start with that, but pls state this, we dont want to merge rough patches to the codebase ;) | 00:21 |
@HeikoS | thoralf: alles klaro? | 00:21 |
thoralf | HeikoS: Remember you promised to care about the documentation when I wrote this Quickstart thing? | 00:22 |
dhruv13 | HeikoS: Sure! | 00:22 |
thoralf | HeikoS: It would be a good time now ;) | 00:22 |
-!- kislay [~kislay@93.114.45.13] has quit [Ping timeout: 240 seconds] | 00:22 | |
@HeikoS | thoralf: ehm? what? | 00:23 |
@HeikoS | haha :) | 00:23 |
@HeikoS | I cannot even remember ;=) | 00:23 |
thoralf | HeikoS: We should use GSoC to lower barriers and have many people who will tell us if something about our intros/quickstarts/tutorials is wrong :) | 00:23 |
@HeikoS | thoralf: I totally agree | 00:24 |
@HeikoS | thoralf: so whats your plan? | 00:24 |
thoralf | HeikoS: Now we have this quickstart (for non-developers), we could include it into the website... | 00:25 |
thoralf | Maybe in a way that it stays up-to-date with github stuff. | 00:26 |
@HeikoS | thoralf: yeah, good! | 00:26 |
thoralf | Any ideas? | 00:26 |
@HeikoS | thoralf: what about a little section: How to contribute | 00:26 |
@HeikoS | thoralf: where this goes, | 00:26 |
@HeikoS | thoralf: along with the points from the gsoc howto | 00:26 |
@HeikoS | and then also some more technical things...how to set up things | 00:26 |
@HeikoS | thoralf: I think we can render github .md files on the website, sören told me about that | 00:27 |
@HeikoS | thoralf: maybe ask kevin about that, he like hacking the website | 00:27 |
thoralf | picklexxx? | 00:27 |
@HeikoS | thoralf: yeah | 00:29 |
@HeikoS | thoralf: I would do this, but I am too busy reviewing code and organising things ... so pls push it, I love the itea | 00:30 |
@HeikoS | thoralf: its just a matter of telling someone what to do ;) | 00:30 |
thoralf | HeikoS: Yeah. | 00:30 |
thoralf | So what about rendering $page.md from github (latest release, not neccessarily master) and put it on the web server? Or do we simply want to link it? | 00:31 |
@HeikoS | thoralf: rendering would be cool I guess | 00:32 |
@HeikoS | if it takes the latest github version | 00:32 |
thoralf | Okay, I'll ask Kevin. | 00:33 |
pfm_ | dhruv13: sorry, had to get some food... it does not go above 32% | 00:34 |
thoralf | Only 32% of food? That's bad. Means you're still hungry? | 00:36 |
pfm_ | dhruv13: currently failing after building shogun-unit-test | 00:36 |
pfm_ | thoralf: haha, i had 100% of the food | 00:36 |
@HeikoS | haha | 00:37 |
@HeikoS | pfm_: whats the problem? | 00:37 |
pfm_ | HeikoS: i'm trying to build/install/compile/words shogun on os x | 00:38 |
pfm_ | and i'm getting errors | 00:38 |
@HeikoS | pfm_: ah osx. I can't help there unfortunately | 00:38 |
@HeikoS | pfm_: We would appreciate it *a lot* if you (once you managed how to do this) describe all problems and the procedure in a README and send a patch | 00:39 |
-!- dhruv13 [~dhruv13@84.64.13.181] has quit [Ping timeout: 244 seconds] | 00:40 | |
pfm_ | HeikoS: i am a bit of a noob so i'm not sure if i can do that, but i can keep a record of what i'm trying | 00:41 |
pfm_ | 'turning it off and on' | 00:41 |
@HeikoS | pfm_: just in case :) | 00:44 |
@HeikoS | pfm_: if it keeps on not working, ask on the mailing list | 00:44 |
@HeikoS | then everyone will read | 00:44 |
pfm_ | kk, i feel like i am making incremental progress so hopefully it'll work | 00:44 |
-!- dhruv13 [~dhruv13@84.64.13.181] has joined #shogun | 00:49 | |
@HeikoS | pfm_: cool! | 00:49 |
@HeikoS | pfm_: again, any tutorial how to build Shogun on macosx is highly appreciated | 00:50 |
@HeikoS | pfm_: just talked to thoralf about putting such things onto our website | 00:50 |
dhruv13 | pfm_: hey! sorry, got disconnected | 00:50 |
pfm_ | dhruv13: i am pretty much still stuck at the unit-test part | 00:51 |
dhruv13 | pfm_: still 32%? did you resolve it? | 00:51 |
pfm_ | no resolution, always 32% | 00:51 |
dhruv13 | pfm_: what build errors are you getting exactly? | 00:52 |
dhruv13 | pfm_: it just quits? | 00:52 |
pfm_ | dhruv13: most recent error is the sme as this one: https://www.dropbox.com/s/kh02askyd5ul2hx/Screenshot%202014-03-06%2017.00.05.png | 00:52 |
pfm_ | 'Error 2' | 00:52 |
pfm_ | descriptive :p | 00:52 |
pfm_ | when it finishes its current business i'm gonna clean it out and start again, in case that helps at all... | 00:53 |
dhruv13 | hmm... there seems to be a config error of some sort... | 00:55 |
dhruv13 | see this: http://stackoverflow.com/questions/5390081/make-gcc-cryptic-error-2-how-to-have-more-information | 00:55 |
dhruv13 | I am not familiar with cmake, but the answer above suggests that there should be some way to make the error output more verbose... | 00:56 |
dhruv13 | does anyone here know how to do that? | 00:57 |
thoralf | pfm_: If you use "make -jxxx", then the errors that cause make to fail may occur very early in the process. | 00:57 |
thoralf | pfm_: Did you use -j? | 00:57 |
pfm_ | thoralf: not this time, but earlier was was using -j3 | 00:57 |
pfm_ | *i was | 00:57 |
thoralf | Hmm. | 00:57 |
thoralf | make VERBOSE=1? | 00:59 |
dhruv13 | thoralf: is there a known conflict between homebrew python bindings and the installation from source on a mac? | 01:00 |
@HeikoS | thoralf: just saw the mail. great! | 01:01 |
pfm_ | https://gist.github.com/corcra/dbaae0934c4c254e516c most recent error | 01:01 |
thoralf | dhruv13: Not as I remember, but I'm the wrong guy for that... ;) | 01:01 |
pfm_ | i thought i solved this by including the explicit swig path in my cmake command though... | 01:01 |
thoralf | HeikoS: Yes, last thing before going to bed. | 01:02 |
thoralf | See ya! | 01:02 |
-!- thoralf [~thoralf@p4FEC8B68.dip0.t-ipconnect.de] has quit [Quit: Konversation terminated!] | 01:03 | |
dhruv13 | pfm_: the problem isn't with swig, it is with a file called swig_typemaps.i which is supposed to be in /src/interfaces/python_modular | 01:06 |
dhruv13 | can you go to your source directory and check if the file is there? | 01:06 |
pfm_ | dhruv13: it is there | 01:09 |
dhruv13 | umm, can you try sudo make -j3? | 01:10 |
dhruv13 | is it possible that there is a permissions problem? | 01:10 |
pfm_ | omg the progress bars are changing | 01:10 |
dhruv13 | ? | 01:11 |
pfm_ | after sudo make -j3 | 01:11 |
pfm_ | i am ... not sure why or what | 01:11 |
dhruv13 | >32% | 01:11 |
pfm_ | i just deleted everything and tried doing ti again | 01:11 |
pfm_ | yep | 01:11 |
pfm_ | 46% atm | 01:11 |
pfm_ | if it works now i will be happy and confused | 01:11 |
pfm_ | cause i am not sure what i changed | 01:12 |
dhruv13 | cool! ithink it needed to create and edit new files, for which it did not have requisite permissions | 01:12 |
dhruv13 | on linux-mint, i have to use sudo before everything, so it didn't strike me earlier | 01:13 |
pfm_ | i should have tried it | 01:13 |
pfm_ | it's usually a fall back :p | 01:13 |
pfm_ | https://xkcd.com/149/ etc. | 01:13 |
dhruv13 | ha ha | 01:14 |
dhruv13 | xkcd is awesome!! | 01:14 |
dhruv13 | never ceases to amuse me | 01:14 |
pfm_ | yeah it's impressive that he keeps it so consistently good | 01:14 |
dhruv13 | yep! ;-) | 01:14 |
dhruv13 | pfm_: progressing right? | 01:19 |
pfm_ | yes, it's finished!! | 01:19 |
dhruv13 | pfm_: cool!! | 01:19 |
pfm_ | i think i need to get python to know where it is though | 01:20 |
dhruv13 | just do sudo make install ;-) | 01:20 |
pfm_ | aha | 01:20 |
pfm_ | yes :p | 01:20 |
dhruv13 | then, python will know... | 01:20 |
pfm_ | i am getting ahead of myself | 01:20 |
dhruv13 | XD | 01:21 |
dhruv13 | anyway, good night!! i think i'll sleep now... | 01:22 |
pfm_ | thanks so much for all the help!! :) | 01:22 |
dhruv13 | no problem! :-) | 01:22 |
dhruv13 | bye! | 01:22 |
-!- dhruv13 [~dhruv13@84.64.13.181] has quit [Quit: Leaving] | 01:22 | |
-!- sunil1337 [0e8bd057@gateway/web/freenode/ip.14.139.208.87] has left #shogun [] | 01:40 | |
-!- HeikoS [~heiko@p4FED4101.dip0.t-ipconnect.de] has quit [Ping timeout: 240 seconds] | 01:59 | |
-!- sirotka [~sirotka@194.1.184.161] has joined #shogun | 02:01 | |
-!- sirotka [~sirotka@194.1.184.161] has quit [Remote host closed the connection] | 02:17 | |
-!- cameo54321 [27298a20@gateway/web/freenode/ip.39.41.138.32] has joined #shogun | 03:04 | |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 03:07 | |
shogun-buildbot | build #735 of nightly_default is complete: Failure [failed doc] Build details are at http://buildbot.shogun-toolbox.org/builders/nightly_default/builds/735 | 03:14 |
-!- shogun-buildbot_ [~shogun-bu@7nn.de] has joined #shogun | 03:21 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Ping timeout: 265 seconds] | 03:25 | |
-!- zxtx_ [~zv@129-79-241-148.dhcp-bl.indiana.edu] has joined #shogun | 04:05 | |
-!- zxtx [~zv@129-79-241-148.dhcp-bl.indiana.edu] has quit [Ping timeout: 244 seconds] | 04:06 | |
-!- zxtx_ is now known as zxtx | 04:13 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 04:23 | |
-!- zxtx [~zv@129-79-241-148.dhcp-bl.indiana.edu] has quit [Ping timeout: 240 seconds] | 04:23 | |
-!- Netsplit *.net <-> *.split quits: shogun-buildbot_ | 04:25 | |
-!- cameo54321 [27298a20@gateway/web/freenode/ip.39.41.138.32] has quit [Ping timeout: 245 seconds] | 06:09 | |
-!- kislay [~kislay@93.114.45.13] has joined #shogun | 06:17 | |
-!- PIDJE [b639e0b6@gateway/web/freenode/ip.182.57.224.182] has joined #shogun | 08:06 | |
-!- sonne|osx [~sonne@89.204.135.86] has joined #shogun | 08:11 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Remote host closed the connection] | 08:13 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 08:13 | |
-!- sonne|osx_ [~sonne@89.204.137.123] has joined #shogun | 08:25 | |
sonne|osx_ | shogun-buildbot: force build --branch=develop 'deb3 - modular interfaces' | 08:26 |
shogun-buildbot | no such builder 'deb3 - modular interfaces' | 08:26 |
-!- sonne|osx [~sonne@89.204.135.86] has quit [Ping timeout: 264 seconds] | 08:26 | |
-!- sonne|osx_ is now known as sonne|osx | 08:26 | |
sonne|osx | shogun-buildbot: force build --branch=develop 'deb3 - modular_interfaces' | 08:26 |
shogun-buildbot | The build has been queued, I'll give a shout when it starts | 08:27 |
sonne|osx | shogun-buildbot: force build --branch=develop 'deb2 - static_interfaces' | 08:27 |
shogun-buildbot | The build has been queued, I'll give a shout when it starts | 08:27 |
sonne|osx | shogun-buildbot: force build --branch=develop 'cyg1 - libshogun' | 08:27 |
shogun-buildbot | build #1857 forced | 08:27 |
shogun-buildbot | I'll give a shout when the build finishes | 08:27 |
sonne|osx | shogun-buildbot: force build --branch=develop 'deb4 - python3' | 08:27 |
shogun-buildbot | The build has been queued, I'll give a shout when it starts | 08:27 |
-!- sonne|osx [~sonne@89.204.137.123] has quit [Quit: sonne|osx] | 08:33 | |
-!- kislay [~kislay@93.114.45.13] has quit [Read error: Connection reset by peer] | 08:38 | |
-!- witness___ [uid10044@gateway/web/irccloud.com/session] has joined #shogun | 08:53 | |
-!- zxtx [~zv@c-98-223-196-32.hsd1.in.comcast.net] has joined #shogun | 08:53 | |
-!- witness___ [uid10044@gateway/web/irccloud.com/session] has quit [Changing host] | 08:53 | |
-!- witness___ [uid10044@gateway/web/irccloud.com/x-fiqqgwbiwdbiohan] has joined #shogun | 08:53 | |
-!- dhruv13 [0e8b7204@gateway/web/freenode/ip.14.139.114.4] has joined #shogun | 08:56 | |
PIDJE | hi | 09:01 |
-!- PIDJE [b639e0b6@gateway/web/freenode/ip.182.57.224.182] has left #shogun [] | 09:02 | |
-!- dhruv13 [0e8b7204@gateway/web/freenode/ip.14.139.114.4] has left #shogun [] | 09:10 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Quit: buildmaster reconfigured: bot disconnecting] | 09:13 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 09:14 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Client Quit] | 09:14 | |
-!- shogun-buildbot_ [~shogun-bu@7nn.de] has joined #shogun | 09:14 | |
shogun-buildbot_ | build #209 of debian wheezy - memcheck is complete: Failure [failed memory check generate html] Build details are at http://buildbot.shogun-toolbox.org/builders/debian%20wheezy%20-%20memcheck/builds/209 blamelist: Parijat Mazumdar <mazumdarparijat@gmail.com>, Orion Poplawski <orion@nwra.com>, Soeren Sonnenburg <sonne@debian.org>, iglesias | 09:50 |
shogun-buildbot_ | <fernando.iglesiasg@gmail.com>, Bj?rn Esser <bjoern.esser@gmail.com>, Thoralf Klein <thoralf.klein@zib.de>, lambday <heavensdevil6909@gmail.com>, Viktor Gal <vigsterkr@gmail.com>, Heiko Strathmann <heiko.strathmann@gmail.com>, tklein23 <tklein23@users.noreply.github.com>, Viktor Gal <viktor.gal@maeth.com>, Thoralf Klein <thoralf@fischlustig.de>, Sergey Lisitsyn | 09:50 |
shogun-buildbot_ | <lisitsyn.s.o@gmail.com>, Fernando Iglesias <fernando.iglesiasg@gmail.com>, Roman Votyakov <votjakovr@gmail.com> | 09:50 |
@wiking | shogun-buildbot_: force build --branch=develop 'debian wheezy - memcheck' | 10:09 |
shogun-buildbot_ | The build has been queued, I'll give a shout when it starts | 10:09 |
-!- lambday [67157f54@gateway/web/freenode/ip.103.21.127.84] has joined #shogun | 10:16 | |
-!- kislay [~kislay@93.114.45.13] has joined #shogun | 10:31 | |
-!- kislay [~kislay@93.114.45.13] has quit [Ping timeout: 265 seconds] | 10:36 | |
-!- HeikoS1 [~heiko@p4FED43A8.dip0.t-ipconnect.de] has joined #shogun | 10:40 | |
-!- kislay [~kislay@93.114.45.13] has joined #shogun | 10:41 | |
-!- HeikoS1 [~heiko@p4FED43A8.dip0.t-ipconnect.de] has quit [Ping timeout: 240 seconds] | 10:46 | |
-!- HeikoS1 [~heiko@p4FED43A8.dip0.t-ipconnect.de] has joined #shogun | 10:56 | |
@wiking | HeikoS1: u in .de? :) | 10:59 |
-!- HeikoS1 [~heiko@p4FED43A8.dip0.t-ipconnect.de] has quit [Ping timeout: 244 seconds] | 11:01 | |
-!- kislay [~kislay@93.114.45.13] has quit [Read error: Connection reset by peer] | 11:14 | |
shogun-buildbot_ | build #210 forced | 11:18 |
shogun-buildbot_ | I'll give a shout when the build finishes | 11:18 |
-!- kislay [~kislay@93.114.45.13] has joined #shogun | 11:29 | |
@wiking | sonney2k: ping | 11:30 |
-!- pirosb3 [8da3e8e0@gateway/web/freenode/ip.141.163.232.224] has joined #shogun | 11:32 | |
pirosb3 | Good morning :) | 11:33 |
pirosb3 | hi besser82 sonney2k wiking | 11:34 |
pirosb3 | HI pfm_ did you manage to solve the issues? | 11:34 |
pirosb3 | PS: anyone available for a quick arch question? | 11:34 |
pirosb3 | http://pastebin.instantbird.com/622462 | 11:35 |
-!- kislay [~kislay@93.114.45.13] has quit [Ping timeout: 264 seconds] | 11:35 | |
-!- HeikoS1 [~heiko@p4FED43A8.dip0.t-ipconnect.de] has joined #shogun | 11:40 | |
@wiking | pirosb3: have u maybe used gcc/g++ for compiling shogun? as for me it seems like this... "___gxx_personality_v0" | 11:41 |
pirosb3 | wiking: it all works now, I had to do -cxx. thanks ayways | 11:42 |
-!- HeikoS1 [~heiko@p4FED43A8.dip0.t-ipconnect.de] has quit [Client Quit] | 11:43 | |
@wiking | pirosb3: ah u mean clang++ ? | 11:43 |
-!- HeikoS1 [~heiko@p4FED43A8.dip0.t-ipconnect.de] has joined #shogun | 11:43 | |
pirosb3 | yes | 11:44 |
-!- lambday_ [67157f54@gateway/web/freenode/ip.103.21.127.84] has joined #shogun | 11:57 | |
pirosb3 | hey guys, can I start on this? https://github.com/shogun-toolbox/shogun/issues/1868 | 12:00 |
-!- lambday [67157f54@gateway/web/freenode/ip.103.21.127.84] has quit [Ping timeout: 245 seconds] | 12:00 | |
HeikoS1 | wiking: nice with the coverage! | 12:01 |
HeikoS1 | pirosb3: hi | 12:01 |
pirosb3 | hi HeikoS1 | 12:01 |
pirosb3 | can I start working on it? https://github.com/shogun-toolbox/shogun/issues/1868 | 12:02 |
HeikoS1 | pirosb3: yeah feel free to make something useful there | 12:02 |
pirosb3 | HeikoS1: what type of format should the data be in? | 12:03 |
pirosb3 | is it up to me? or do you suggest something? | 12:03 |
HeikoS1 | pirosb3: you could serialise the model | 12:05 |
HeikoS1 | pirosb3: or use something similar to LibSVM | 12:05 |
HeikoS1 | pirosb3: but ask thoralf in the issue | 12:05 |
HeikoS1 | pirosb3: any shoun class can be serialised trivially | 12:05 |
HeikoS1 | pirosb3: you just need to make sure the data is not stored again, but just the model. | 12:05 |
pirosb3 | thanks | 12:07 |
pirosb3 | HeikoS1: I am having problems compiling the Python interface | 12:07 |
pirosb3 | how should this be done? | 12:07 |
pirosb3 | I am inside the folder shogun/build-osx/src/interfaces/python_modular | 12:07 |
pirosb3 | and I run make, but it fails. Is this the correct way to build modshogun? | 12:08 |
-!- lisitsyn1 [~lisitsin@mxs.kg.ru] has joined #shogun | 12:08 | |
HeikoS1 | you just have to activate it in cmalke | 12:09 |
HeikoS1 | cmake | 12:09 |
HeikoS1 | -DPythonModular=ON | 12:09 |
HeikoS1 | then a std "make" does the job | 12:09 |
pirosb3 | already made it | 12:10 |
pirosb3 | in fact it starts compiling, but fails http://pastebin.instantbird.com/622543 | 12:11 |
-!- kislay [~kislay@93.114.45.13] has joined #shogun | 12:11 | |
HeikoS1 | pirosb3: wow, never seen that error | 12:13 |
HeikoS1 | pirosb3: send a bug report to the mailing list, include all your build details and logs | 12:13 |
HeikoS1 | pirosb3: I am sure people can help you there | 12:13 |
pirosb3 | Thanks | 12:13 |
pirosb3 | for now I am using a VM to do the issue | 12:14 |
@wiking | HeikoS1: ossx related... | 12:14 |
@wiking | HeikoS1: we need $$ for osx buildbot | 12:14 |
HeikoS1 | wiking: yeah the great unknown ;) | 12:14 |
HeikoS1 | wiking: yeah we need | 12:14 |
HeikoS1 | wiking: mmh, maybe after this years gsoc we can afford one | 12:15 |
HeikoS1 | or we have a donation button on the page | 12:15 |
pirosb3 | HeikoS1: I had already filed an issue https://github.com/shogun-toolbox/shogun/issues/1925 | 12:15 |
-!- lambday [67157e54@gateway/web/freenode/ip.103.21.126.84] has joined #shogun | 12:20 | |
-!- lambday_ [67157f54@gateway/web/freenode/ip.103.21.127.84] has quit [Ping timeout: 245 seconds] | 12:22 | |
pirosb3 | HeikoS1: what if we use svmlight format to train the classifier? | 12:24 |
pirosb3 | in this way, we only give one file (instead of 2 files: train and labels) | 12:24 |
HeikoS1 | pirosb3: maybe thats a good idea | 12:24 |
pirosb3 | and provide a standard that is more easily used | 12:24 |
HeikoS1 | pirosb3: I think we can already read/write that | 12:24 |
pirosb3 | I will write it on the ticket+ | 12:24 |
HeikoS1 | pirosb3: but it would be a little annyoing to write a parser | 12:25 |
pirosb3 | yes, but an automated tool is great | 12:25 |
pirosb3 | mmm.... | 12:25 |
pirosb3 | <line> .=. <target> <feature>:<value> <feature>:<value> ... <feature>:<value> # | 12:25 |
HeikoS1 | pirosb3: so you could even just serialise, as things would work out of the box | 12:25 |
pirosb3 | HeikoS1: serialize the classifier, or the dataset? | 12:25 |
pirosb3 | the classifier right? | 12:25 |
HeikoS1 | pirosb3: yeah classifier without data | 12:26 |
HeikoS1 | pirosb3: so you would need to detach everything that is not the classifier itself | 12:26 |
pirosb3 | In my idea, the tool would be like this: | 12:27 |
HeikoS1 | pirosb3: see for example store_model_features in CSVM | 12:27 |
pirosb3 | ok | 12:28 |
pirosb3 | python train_multiclass_svm.py train.svmlight multiclass-svm.serialized-model | 12:29 |
pirosb3 | mmmm... in python predict_multiclass_svm.py multiclass-svm.model eval.data predicted.labels | 12:30 |
pirosb3 | is predicted.labels the output? | 12:30 |
HeikoS1 | pirosb3: predicted label is output on test data I guess | 12:31 |
HeikoS1 | pirosb3: discuss this with thoralf pls. I have to go now, good luck! | 12:31 |
pirosb3 | Thanks HeikoS1 I will update the ticket | 12:32 |
-!- HeikoS1 [~heiko@p4FED43A8.dip0.t-ipconnect.de] has quit [Quit: Leaving.] | 12:36 | |
-!- HeikoS1 [~heiko@p4FED43A8.dip0.t-ipconnect.de] has joined #shogun | 12:36 | |
-!- HeikoS1 [~heiko@p4FED43A8.dip0.t-ipconnect.de] has quit [Ping timeout: 264 seconds] | 12:44 | |
-!- pirosb3 [8da3e8e0@gateway/web/freenode/ip.141.163.232.224] has quit [Quit: Page closed] | 13:04 | |
-!- kislay [~kislay@93.114.45.13] has quit [Read error: Connection reset by peer] | 13:04 | |
-!- PirosB3_ [~pirosb3@162.219.6.191] has joined #shogun | 13:06 | |
-!- lambday [67157e54@gateway/web/freenode/ip.103.21.126.84] has quit [Ping timeout: 245 seconds] | 13:10 | |
-!- prams [~prams@117.237.140.54] has joined #shogun | 13:12 | |
shogun-buildbot_ | build #210 of debian wheezy - memcheck is complete: Failure [failed memory check] Build details are at http://buildbot.shogun-toolbox.org/builders/debian%20wheezy%20-%20memcheck/builds/210 | 13:17 |
-!- kislay [~kislay@93.114.45.13] has joined #shogun | 13:21 | |
-!- prams [~prams@117.237.140.54] has quit [Read error: Connection reset by peer] | 13:24 | |
-!- kislay [~kislay@93.114.45.13] has quit [Read error: Connection reset by peer] | 13:28 | |
-!- lambday [67157d54@gateway/web/freenode/ip.103.21.125.84] has joined #shogun | 13:31 | |
-!- cameo54321 [27294156@gateway/web/freenode/ip.39.41.65.86] has joined #shogun | 13:37 | |
-!- lambday [67157d54@gateway/web/freenode/ip.103.21.125.84] has quit [Ping timeout: 245 seconds] | 13:41 | |
-!- kislay [~kislay@93.114.45.13] has joined #shogun | 13:44 | |
-!- cameo54321 [27294156@gateway/web/freenode/ip.39.41.65.86] has quit [Ping timeout: 245 seconds] | 14:02 | |
-!- Saurabh7 [~Saurabh7@115.248.130.148] has joined #shogun | 14:18 | |
PirosB3_ | hey guys | 14:26 |
PirosB3_ | can I have some early feedback on this? https://github.com/PirosB3/shogun/commit/b7158eccd2c1f76529b02a148a0f896c03c84c31 (related to https://github.com/shogun-toolbox/shogun/issues/1868) | 14:26 |
PirosB3_ | thanks? | 14:26 |
PirosB3_ | oops, I meant thanks! | 14:26 |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 14:27 | |
shogun-notifier- | shogun: Viktor Gal :develop * 3ca1191 / tests/unit/base/main_unittest.cc: https://github.com/shogun-toolbox/shogun/commit/3ca11915acb0d482a6ddc0f90de5d83ef970ad93 | 14:27 |
shogun-notifier- | shogun: revert the changes of 56ceab6 for main_unittest.cc | 14:27 |
@wiking | PirosB3_: send in a PR and somebody will review it... | 14:28 |
@wiking | PR = pull request | 14:29 |
PirosB3_ | yep, but should I do it now or when I'm finished?^ | 14:30 |
@wiking | PirosB3_: do not ask people to review things in your own repo | 14:31 |
PirosB3_ | ok | 14:31 |
PirosB3_ | so I should pull request, and then ask for a review | 14:31 |
@wiking | PirosB3_: dont have to ask for a RP | 14:32 |
@wiking | i mean review | 14:32 |
@wiking | somebody will review it if u send a PR | 14:32 |
PirosB3_ | ok | 14:32 |
-!- witness___ [uid10044@gateway/web/irccloud.com/x-fiqqgwbiwdbiohan] has quit [Quit: Connection closed for inactivity] | 14:34 | |
-!- HeikoS1 [~heiko@p4FED43A8.dip0.t-ipconnect.de] has joined #shogun | 14:45 | |
PirosB3_ | wiking: I sent a PR | 14:53 |
PirosB3_ | anyone: I get MulticlassStrategy::CMulticlassStrategy(): register parameters! can anyone give me more details on why this happens? thanks | 14:54 |
@wiking | PirosB3_: ok somebody will get back to u with some comments for sure | 14:55 |
PirosB3_ | all I can see on MulticlassOneVsOneStrategy is that there is no real condition to send this warning | 14:57 |
-!- ckkay [~ckkay@chello080108038237.1.11.vie.surfer.at] has joined #shogun | 14:57 | |
kislay | wiking, is there some way we can run uint8 datatype Bytefeatures in classes like EuclideanDistance(), PCA(). | 14:58 |
kislay | Converting uint8 to double for using these classes are prone to quite amount of stray errors. | 14:58 |
PirosB3_ | ok going 4 lunch, hear u later! | 14:59 |
@wiking | kislay: i'm afraid PCA works only on float64_t :< | 15:00 |
kislay | hmm.. I have quite satisfactorily implemented eigenfaces for showing its compression standards. | 15:01 |
@wiking | kislay: there's a fundamental problem with the preprocessors, i.e. they are not being templatized like Features with datatype... and most of them only works with float64_t | 15:01 |
kislay | but showing eigenfaces for face recognition requires more accuracy | 15:01 |
@wiking | of course one reason for this is that most of the preprocessors will need floating point numbers... but still this should b handled somehow internally | 15:02 |
@wiking | kislay: why dont u just convert your int8 to float | 15:03 |
@wiking | if u use opencv | 15:03 |
@wiking | that's like 1liner | 15:03 |
@wiking | or mostly with any of the good image processing libs that's a 1 liner | 15:03 |
kislay | I am doing that only . It works nice for showing pca's dimensional reductionality, but when i use that data for correlation between the test and the training image...things get weird :( | 15:04 |
-!- HeikoS1 [~heiko@p4FED43A8.dip0.t-ipconnect.de] has quit [Quit: Leaving.] | 15:05 | |
kislay | I am using eucledian distance for matching test image and training image. | 15:05 |
-!- HeikoS1 [~heiko@p4FED43A8.dip0.t-ipconnect.de] has joined #shogun | 15:05 | |
kislay | some suggestions of using any other distance measure? | 15:05 |
-!- HeikoS1 [~heiko@p4FED43A8.dip0.t-ipconnect.de] has quit [Client Quit] | 15:07 | |
-!- HeikoS1 [~heiko@p4FED43A8.dip0.t-ipconnect.de] has joined #shogun | 15:07 | |
@wiking | kislay: well how do u represent the image? | 15:08 |
shogun-buildbot_ | build #476 of precise - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/precise%20-%20libshogun/builds/476 blamelist: Viktor Gal <viktor.gal@maeth.com> | 15:09 |
shogun-buildbot_ | build #520 of FC19 - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/FC19%20-%20libshogun/builds/520 blamelist: Viktor Gal <viktor.gal@maeth.com> | 15:09 |
kislay | flattened row vectors | 15:09 |
@wiking | kislay: pixel-by-pixel? | 15:10 |
kislay | yeah! | 15:10 |
@wiking | pfff | 15:10 |
-!- prams [~prams@117.235.198.247] has joined #shogun | 15:10 | |
@wiking | that's not a good way... | 15:10 |
kislay | oh! | 15:11 |
kislay | what's the better way then? | 15:11 |
@wiking | well you should somehow use some sort of higher level represenation of the image | 15:12 |
@wiking | kislay: there's tons of articles about this | 15:12 |
kislay | oh! | 15:13 |
kislay | Actually I think I answered you wrong | 15:14 |
kislay | I condense say 50 , 100X100 image to be generated from different combinations of 20 eigenfaces(again 100X100 images). | 15:15 |
kislay | so i think, i am doing what you are pointing at. | 15:16 |
-!- sunil1337 [0e8bd057@gateway/web/freenode/ip.14.139.208.87] has joined #shogun | 15:22 | |
@wiking | HeikoS1: how do u like this one: https://github.com/shogun-toolbox/shogun/issues/1955 | 15:29 |
-!- lisitsyn1 [~lisitsin@mxs.kg.ru] has quit [Quit: Leaving.] | 15:31 | |
-!- dhruv13 [~dhruv13@84.64.14.14] has joined #shogun | 15:31 | |
HeikoS1 | wiking: I like it | 15:36 |
dhruv13 | wiking: regarding your suggestion here https://github.com/shogun-toolbox/shogun/issues/1936, won't using nanoflann increase dependencies? | 15:39 |
dhruv13 | wiking: or do you want me to use them as a guideline? | 15:39 |
sunil1337 | HeikoS1: build fails.. error:SSE2 instruction set not enabled... it is fixed by adding -msse -msse2 -msse3 in the cmakelist.txt file.. do i commit it? | 15:40 |
@wiking | HeikoS1: no | 15:40 |
@wiking | err | 15:40 |
@wiking | sunil1337: no | 15:40 |
sunil1337 | @wiking: okay.... | 15:41 |
@wiking | sunil1337: show us why that happens | 15:41 |
HeikoS1 | wiking: hey, triggering the buildbot to run memchecks? | 15:41 |
@wiking | pastebin or whatever | 15:41 |
HeikoS1 | wiking: or is that overkill? | 15:41 |
@wiking | HeikoS1: could work... but man it really takes ages to run those | 15:41 |
@wiking | HeikoS1: get us a new machine... | 15:41 |
@wiking | :) | 15:42 |
@wiking | or another buildbot | 15:42 |
HeikoS1 | wiking: hehe | 15:42 |
HeikoS1 | wiking: talking to deepmind soon hopefully :) | 15:42 |
sunil1337 | @wiking: it happens when creating object for random.cpp in mathematics when <emmintrin.h> is included | 15:42 |
@wiking | dhruv13: well it all depends... nanoflann should be reviewed how much dependency is being there added above nanoflann and discuss it in the issue and then we can do an educated decision | 15:42 |
HeikoS1 | wiking: ok the leave that, I will ask everyone in the PR whether valgrind was ran | 15:42 |
HeikoS1 | before I merge | 15:42 |
@wiking | HeikoS1: you mean google? :D | 15:42 |
HeikoS1 | wiking: yep! :D | 15:43 |
@wiking | sunil1337: yeah i was guessing so | 15:43 |
HeikoS1 | wiking: they sponsor some stuff at UCL, so I am optimistic | 15:43 |
HeikoS1 | wiking: but need to sort out some things first. and currently all this screening students takes lots of tiume | 15:43 |
HeikoS1 | wiking: though very useful | 15:43 |
sunil1337 | @wiking: is it just for me..or for everyone.... | 15:43 |
dhruv13 | wiking: okay... will do that :-) | 15:43 |
@wiking | sunil1337: well u r the first one obviously otherwise it would be already fixed... | 15:44 |
HeikoS1 | wiking: any news on notebook thing? :) | 15:44 |
@wiking | HeikoS1: will u pay me? :) | 15:44 |
@wiking | HeikoS1: sorry but i have to work as well... | 15:44 |
sunil1337 | @wiking: okay... lets wait for others then :).. | 15:44 |
@wiking | HeikoS1: but i'm trying my best | 15:45 |
HeikoS1 | wiking: mmmmmh, ..... | 15:45 |
HeikoS1 | wiking: no | 15:45 |
dhruv13 | wiking: might be a useful update to the KNN techniques if nanoflann is used; PCL uses flann for their knn search... | 15:45 |
HeikoS1 | :) | 15:45 |
@wiking | sunil1337: send in an issue | 15:45 |
sunil1337 | @wiking: okay! | 15:45 |
@wiking | sunil1337: with cmake output + error | 15:45 |
HeikoS1 | wiking: I need the webdemos to work and to be cool to convince people, and the shogun missionary poroject being done. Then we are ready to go | 15:45 |
@wiking | do not interpret the output please... copy-paste it as is | 15:45 |
HeikoS1 | wiking: ok thanks, I just want to make sure its on your mind :) | 15:45 |
dhruv13 | wiking: and yeah, sorry about the mess with main_unittest.cc | 15:45 |
@wiking | HeikoS1: yeah it is | 15:45 |
@wiking | dhruv13: nw | 15:45 |
dhruv13 | wiking: the commented out lines were to help noobs like me debug unit test errors.. | 15:46 |
@wiking | dhruv13: trust me i got that before saying it the 3rd time ;) | 15:49 |
@wiking | </sarcasm> | 15:49 |
dhruv13 | lol | 15:49 |
-!- HeikoS1 [~heiko@p4FED43A8.dip0.t-ipconnect.de] has quit [Ping timeout: 240 seconds] | 15:50 | |
@wiking | dhruv13: for this reason i've just issued https://github.com/shogun-toolbox/shogun/issues/1955 | 15:50 |
@wiking | so that one can change log level via env variable.. | 15:50 |
dhruv13 | cool! will take a look at that ;-) | 15:50 |
-!- dhruv13 [~dhruv13@84.64.14.14] has quit [Quit: Leaving] | 15:51 | |
-!- sunil1337 [0e8bd057@gateway/web/freenode/ip.14.139.208.87] has quit [Quit: Page closed] | 15:56 | |
-!- tomy18 [~alex@79.114.164.19] has joined #shogun | 15:56 | |
-!- sunil1337 [0e8bd057@gateway/web/freenode/ip.14.139.208.87] has joined #shogun | 16:06 | |
sunil1337 | @wiking: strange! no errors now.. | 16:06 |
sunil1337 | @wiking: anyways thanks.! | 16:07 |
-!- sunil1337 [0e8bd057@gateway/web/freenode/ip.14.139.208.87] has quit [Client Quit] | 16:08 | |
-!- HeikoS1 [~heiko@p4FED43A8.dip0.t-ipconnect.de] has joined #shogun | 16:13 | |
-!- HeikoS2 [~heiko@p4FED43A8.dip0.t-ipconnect.de] has joined #shogun | 16:18 | |
-!- HeikoS1 [~heiko@p4FED43A8.dip0.t-ipconnect.de] has quit [Ping timeout: 240 seconds] | 16:21 | |
-!- kislay [~kislay@93.114.45.13] has quit [Read error: Connection reset by peer] | 16:21 | |
PirosB3_ | hey guys, why does Shogun Cloud give me loads of 404s? | 16:22 |
PirosB3_ | is it something you are aware of? | 16:22 |
@wiking | PirosB3_: well docker image needs an update ... | 16:23 |
PirosB3_ | is there a ticket on it? | 16:24 |
PirosB3_ | wiking: can I ask you a very shaming question on Gaussian Kernals? :) | 16:24 |
-!- HeikoS2 [~heiko@p4FED43A8.dip0.t-ipconnect.de] has quit [Ping timeout: 264 seconds] | 16:24 | |
PirosB3_ | why does it accept 2 parameters? | 16:25 |
PirosB3_ | x and x1 | 16:25 |
@wiking | PirosB3_: noup there's no ticket for it as it requires quite amount of work... as i believe trusted builds of docker will never support submodules if they continue like this so we have to setup our own docker image repository + buildbot | 16:25 |
@wiking | PirosB3_: doxygen is your friend: http://shogun-toolbox.org/doc/en/3.0.0/classshogun_1_1CGaussianKernel.html | 16:26 |
PirosB3_ | wiking: when you speak about submodules, do you mean multiple docker containers connected together? | 16:27 |
@wiking | PirosB3_: i talk about git submodules | 16:27 |
PirosB3_ | I already went there wiking , my question is on Gaussian kernels in general | 16:28 |
PirosB3_ | wiking: you can make a init script that does that | 16:28 |
@wiking | PirosB3_: ? | 16:28 |
@wiking | PirosB3_: have u tried trusted builds of index.docker.io? | 16:28 |
@wiking | with a repo that actually has submodules | 16:28 |
@wiking | ? | 16:28 |
@wiking | please elaborate if u've managed to get it work | 16:28 |
-!- prams [~prams@117.235.198.247] has left #shogun [] | 16:29 | |
@wiking | as currently based on docker developers it is not supported. | 16:29 |
PirosB3_ | no, but I might have some ideas | 16:29 |
@wiking | repo with submodules to be built with their trusted buildsystem | 16:29 |
@wiking | PirosB3_: so unless you have access to their system where you can fix it, there's no way we can fix it ... | 16:30 |
@wiking | with an init sscript... | 16:30 |
PirosB3_ | https://github.com/shogun-toolbox/shogun/blob/develop/configs/Dockerfile | 16:32 |
PirosB3_ | so you are doing submodule init | 16:32 |
PirosB3_ | you could do it with a hack | 16:33 |
PirosB3_ | but it probably is not worth it | 16:33 |
PirosB3_ | I would suggest to use your own build system, and then push to the docker index | 16:34 |
@wiking | PirosB3_: actually line 31-34 should be removed... | 16:34 |
PirosB3_ | my question on Gaussian kernels: why are there left hand side and right hand side | 16:34 |
@lisitsyn | PirosB3_: because kernel is a function of two arguments ;) | 17:21 |
PirosB3_ | lisitsyn: can you explain this? | 17:23 |
@lisitsyn | PirosB3_: linear kernel is just dot product | 17:23 |
@lisitsyn | distance is a measure of dissimilarity | 17:23 |
@lisitsyn | and kernel is a measure of similarity | 17:23 |
PirosB3_ | mmm | 17:24 |
PirosB3_ | wow ok | 17:24 |
PirosB3_ | so I know the gaussian function | 17:24 |
PirosB3_ | and if I have some points in 2d space | 17:24 |
PirosB3_ | I know I can apply the gaussian fn to those points, to make it more smooth | 17:25 |
PirosB3_ | is that correct? | 17:25 |
@lisitsyn | PirosB3_: that's called kernel density estimation | 17:25 |
PirosB3_ | ok | 17:25 |
PirosB3_ | aaah! KDE | 17:25 |
PirosB3_ | so KDE == covering points in Gaussian | 17:26 |
@lisitsyn | PirosB3_: well it could use various functions | 17:26 |
@lisitsyn | well actually the idea is | 17:26 |
@lisitsyn | you have some measures | 17:26 |
@lisitsyn | but you need to interpolate intermediate values | 17:26 |
@lisitsyn | you can smooth them with gaussian kernels or epanechnikov kernel or whatever :D | 17:26 |
@lisitsyn | but that's different kind of kernel | 17:27 |
@lisitsyn | in things like SVMs and gaussian processes | 17:27 |
PirosB3_ | so kernels always have 2 params? | 17:27 |
@lisitsyn | not in case of KDE | 17:27 |
PirosB3_ | sorry for being so noob | 17:27 |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 17:27 | |
PirosB3_ | can you define me what a kernel is? | 17:27 |
@lisitsyn | in case of KDE it has one parameter | 17:27 |
@lisitsyn | for KDE it is f(x) so that it tells you some density at point x | 17:28 |
@lisitsyn | for SVMs it is f(x,y) which tells you how similar x and y are | 17:28 |
-!- sonne|osx [~sonne@89.204.138.127] has joined #shogun | 17:29 | |
PirosB3_ | ok | 17:29 |
PirosB3_ | in some of the code | 17:29 |
PirosB3_ | I find: kernel = GaussianKernel(feats, feats, width) | 17:29 |
PirosB3_ | this is exactly the same as KDE | 17:29 |
PirosB3_ | correct? | 17:29 |
@lisitsyn | no not really | 17:29 |
@lisitsyn | it is just to indicate | 17:29 |
@lisitsyn | you want to compute all possible f(x,y) where both x and y are from some set of vectors | 17:30 |
@lisitsyn | if you have different lhs and rhs | 17:30 |
@lisitsyn | it would restrict possible values with lhs and rhs | 17:30 |
PirosB3_ | ok | 17:30 |
@lisitsyn | when you train an SVM | 17:30 |
PirosB3_ | can I plot this kerel? | 17:30 |
@lisitsyn | you can plot kernel matrix | 17:30 |
PirosB3_ | I want to see the effect it has on it | 17:30 |
PirosB3_ | how can I actually see the different that the gaussian kernel does to the data | 17:31 |
@lisitsyn | PirosB3_: well we had some demo plotting kernel matrix | 17:31 |
@lisitsyn | PirosB3_: but actually | 17:31 |
PirosB3_ | oh really? | 17:31 |
PirosB3_ | on cloud? | 17:31 |
@lisitsyn | yes | 17:31 |
@lisitsyn | no not on cloud | 17:32 |
PirosB3_ | ok | 17:32 |
@lisitsyn | let me find it | 17:32 |
PirosB3_ | so is there a difference between lhs and rhs? could they be swapped? | 17:32 |
PirosB3_ | looks like it | 17:32 |
PirosB3_ | euclidean distance | 17:33 |
@lisitsyn | PirosB3_: http://demos.shogun-toolbox.org/misc/kernel_matrix/ | 17:33 |
PirosB3_ | whooooa! | 17:33 |
PirosB3_ | ok, now I totally do not have a clue! | 17:34 |
@lisitsyn | PirosB3_: they could be swapped but this would produce transposed matrix | 17:34 |
PirosB3_ | this stuff is so cool tho | 17:34 |
@lisitsyn | PirosB3_: when you train an SVM | 17:34 |
@lisitsyn | you need dot products of vectors from the training set | 17:34 |
@lisitsyn | so lhs == rhs in this case | 17:34 |
@lisitsyn | when you want to classify | 17:34 |
@lisitsyn | you need to get dot products of vectors from the training set and the test set | 17:35 |
@lisitsyn | and lhs == train, rhs == test | 17:35 |
@lisitsyn | in this case | 17:35 |
PirosB3_ | aha! ok | 17:35 |
PirosB3_ | but is the kernel_matrix working correctly? | 17:35 |
PirosB3_ | I cannot find any relation to the kernel and the data | 17:35 |
@lisitsyn | plot two pairs of two points | 17:36 |
@lisitsyn | but separate from each other | 17:36 |
@lisitsyn | I mean e.g. 2 around the point (-3,3) | 17:36 |
@lisitsyn | and 2 around the point (3,3) | 17:36 |
@lisitsyn | so what you'd get is | 17:37 |
@lisitsyn | on diagonal you will get red blocks | 17:37 |
PirosB3_ | how do I change color? | 17:37 |
@lisitsyn | it is not possible I guess | 17:37 |
@lisitsyn | why? | 17:38 |
PirosB3_ | ok | 17:38 |
PirosB3_ | so now I have 4 squared | 17:38 |
PirosB3_ | 2 red 2 blue | 17:38 |
@lisitsyn | hah okay | 17:38 |
@lisitsyn | diagonal elements are 1 | 17:39 |
@lisitsyn | so they are red | 17:39 |
@lisitsyn | this is because k(x,x) = 1 | 17:39 |
@lisitsyn | if kernel is normalized | 17:39 |
@lisitsyn | when you get just blue it is like 0 | 17:39 |
@lisitsyn | so they are pretty much far away from each other | 17:39 |
PirosB3_ | so how does this help the SVM? | 17:40 |
PirosB3_ | by generalizing more? | 17:40 |
@lisitsyn | okay | 17:40 |
@lisitsyn | so if you use linear kernel | 17:40 |
@lisitsyn | which is dot product | 17:40 |
@lisitsyn | SVM finds a hyperplane | 17:40 |
@lisitsyn | if you use gaussian kernel | 17:40 |
@lisitsyn | SVM still find a hyperplane but in totally another space | 17:40 |
@lisitsyn | because this kernel trick induces new hilbert space | 17:41 |
@lisitsyn | you just pretend to use some other space by changing the kernel function | 17:41 |
PirosB3_ | i really want to learn this stuff | 17:41 |
PirosB3_ | where do I start from | 17:41 |
PirosB3_ | I am doing the Andrew NG | 17:41 |
@lisitsyn | but you don't compute it at all | 17:41 |
@lisitsyn | the funny thing about gaussian kernel | 17:41 |
@lisitsyn | is that it virtually creates infinite dimensional space for you | 17:42 |
PirosB3_ | how can it do that? | 17:42 |
@lisitsyn | and the hyperplane in this space could be pretty much any surface | 17:42 |
@lisitsyn | PirosB3_: say we want to separate points | 17:42 |
@lisitsyn | 1, 2, 3, 4 | 17:42 |
@lisitsyn | and we know 2 and 3 are from the class +1 | 17:43 |
@lisitsyn | and 1 and 4 are from the class -1 | 17:43 |
PirosB3_ | gaussian | 17:43 |
@lisitsyn | if we add one more feature | 17:43 |
PirosB3_ | in the middle | 17:43 |
@lisitsyn | x, x^2 | 17:43 |
@lisitsyn | it would be | 17:43 |
@lisitsyn | (1,1), (2,4), (3,9), (4, 16) | 17:44 |
PirosB3_ | yep | 17:44 |
@lisitsyn | now we can separate them with a line | 17:44 |
@lisitsyn | right? | 17:44 |
@lisitsyn | so we just defined a kernel | 17:44 |
@lisitsyn | k(x,y) = x*y + x^2*y^2 | 17:44 |
@lisitsyn | once we map back this hyperplane | 17:44 |
PirosB3_ | yes | 17:45 |
@lisitsyn | we get something like interval | 17:45 |
@lisitsyn | it works pretty much like that | 17:45 |
@lisitsyn | gaussian kernel makes any set of points separable | 17:45 |
@lisitsyn | no matter how they are placed in this space | 17:45 |
PirosB3_ | aha! | 17:45 |
@lisitsyn | you always can find a hyperplane which separates them | 17:45 |
@lisitsyn | in this infinite dimensional space | 17:46 |
PirosB3_ | maybe I got it | 17:46 |
@lisitsyn | if you map this thing back what you'd get is various curves on 2d space | 17:46 |
PirosB3_ | where can I learn more? | 17:46 |
@lisitsyn | andrew ng should be pretty much ok :) | 17:46 |
PirosB3_ | ok | 17:47 |
PirosB3_ | where did you learn this stuff? | 17:47 |
@lisitsyn | it was a long way already :D | 17:47 |
PirosB3_ | ahah | 17:47 |
@lisitsyn | I can't say anything specific | 17:47 |
PirosB3_ | yeah I am really fascinated by ML | 17:47 |
PirosB3_ | this is why I am here | 17:47 |
@lisitsyn | I have read quite a few books and so on | 17:47 |
PirosB3_ | the idea is to use SoC to create something useful and fill in the mysterious gaps I need to understand :) | 17:48 |
PirosB3_ | did you see my proposal on the ML? | 17:48 |
@lisitsyn | PirosB3_: yeah | 17:48 |
PirosB3_ | what do you think? | 17:48 |
-!- sonne|osx [~sonne@89.204.138.127] has quit [Quit: sonne|osx] | 17:48 | |
@lisitsyn | PirosB3_: the main thing for us is to get more contributors ;) | 17:48 |
PirosB3_ | :) | 17:49 |
@lisitsyn | PirosB3_: the design part is pretty challenging | 17:49 |
PirosB3_ | I am restructuring the proposal | 17:49 |
PirosB3_ | the idea would be to start from something less challenging for proof of concept | 17:49 |
@lisitsyn | yeah | 17:49 |
@lisitsyn | makes sense | 17:49 |
@lisitsyn | okay I'll be back a bit later | 17:50 |
PirosB3_ | I feel sort of okay on the design (never say never) because I built one for myself | 17:50 |
PirosB3_ | only does Naive Bayes | 17:50 |
@lisitsyn | heh yeah it makes sense | 17:50 |
PirosB3_ | I am using it for real time sentiment analysis | 17:50 |
@lisitsyn | to start with something then enhance | 17:50 |
PirosB3_ | the idea is create a new maintainable one | 17:51 |
pfm_ | PirosB3_: i did! i deleted everything, started again and remembered to use sudo ;p | 17:51 |
PirosB3_ | hey pfm_ !! | 17:51 |
PirosB3_ | I'm happy it worked | 17:51 |
pfm_ | me too :p | 17:51 |
pfm_ | it took N hours to install and about 30 minutes to get binary classification going | 17:52 |
PirosB3_ | welcome to my world pfm_ | 17:53 |
pfm_ | oh yay, my collaborator just sent me multiclass data | 17:54 |
PirosB3_ | nice :) | 17:55 |
PirosB3_ | pfm_: what will you be doing for SoC? | 17:55 |
pfm_ | PirosB3_: oh i'm not doing SoC, i'm a phd student | 17:56 |
PirosB3_ | you can still do SoC | 17:56 |
PirosB3_ | if you wanted 2 | 17:56 |
PirosB3_ | anyways PhD in what? | 17:56 |
pfm_ | i will be coding in the summer but i already have a stipend, so i would not deprive someone else who needs it :p | 17:57 |
pfm_ | phd in computational biology | 17:57 |
PirosB3_ | thanks pfm_ :) | 17:58 |
PirosB3_ | where are you from pfm_ ^ | 17:58 |
PirosB3_ | lisitsyn: could you please tell me why it is always saying: MulticlassOneVsOneStrategy::CMulticlassOneVsOneStrategy(): register parameters! | 17:59 |
PirosB3_ | I was looking at the source code, and there is no reason why it says this, is it just a warning? | 17:59 |
pfm_ | PirosB3_: i am from ireland but i am studying in the USA, what about you? | 18:00 |
PirosB3_ | i'm originally italian | 18:01 |
PirosB3_ | but studying in UK | 18:01 |
-!- cameo54321 [27293c04@gateway/web/freenode/ip.39.41.60.4] has joined #shogun | 18:01 | |
PirosB3_ | hey guys, I'm just going back home. Hear you after :O | 18:02 |
-!- PirosB3_ [~pirosb3@162.219.6.191] has quit [Quit: PirosB3_] | 18:02 | |
-!- HeikoS [~heiko@p4FED43A8.dip0.t-ipconnect.de] has joined #shogun | 18:09 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 18:09 | |
-!- HeikoS [~heiko@p4FED43A8.dip0.t-ipconnect.de] has quit [Read error: Connection reset by peer] | 18:13 | |
-!- HeikoS1 [~heiko@p4FED43A8.dip0.t-ipconnect.de] has joined #shogun | 18:13 | |
cameo54321 | HeikoS1 Hi! You there? | 18:32 |
HeikoS1 | cameo54321: I am about to leave, whats up? | 18:33 |
cameo54321 | Well I have got the boost libraries found by shogun.. Stuck on stan at the moment | 18:34 |
cameo54321 | stan does not have a cmakelists.txt kind of structure | 18:34 |
cameo54321 | it just has a make file.. it doesn't install | 18:34 |
cameo54321 | if we build it from source | 18:35 |
HeikoS1 | cameo54321: yeah it would just to have to be build in the source tree | 18:35 |
cameo54321 | I am talking about the mcmc entrance task here, the one uploaded about 15 days ago | 18:35 |
HeikoS1 | and then linked against from there | 18:35 |
HeikoS1 | yeah I know | 18:35 |
HeikoS1 | cameo54321: maybe start with the probability distribution interface first? | 18:35 |
HeikoS1 | cameo54321: like what methods and classes are needed | 18:35 |
cameo54321 | you mean i should copy the source files into the src folder of shogun and compile? | 18:36 |
HeikoS1 | cameo54321: know what I mean? | 18:36 |
HeikoS1 | cameo54321: no, we can make cmake to download stan into external_libs dir just like colpack/eigen3(if nonavailable), etc | 18:36 |
HeikoS1 | and then do some implementations based on boost and or stan once that is done | 18:36 |
HeikoS1 | cameo54321: whats the issue number? | 18:36 |
HeikoS1 | cameo54321: 1929 | 18:37 |
cameo54321 | the one i am working on now is https://github.com/shogun-toolbox/shogun/issues/1875 | 18:37 |
HeikoS1 | cameo54321: is first about interfaces and base classes | 18:37 |
HeikoS1 | cameo54321: maybe first go for the 1929 one | 18:38 |
HeikoS1 | I reference that | 18:38 |
cameo54321 | I was under the impression that to do 1929 we'll have to do 1875 first | 18:38 |
HeikoS1 | its somewhat easier to start with | 18:38 |
HeikoS1 | nono | 18:38 |
HeikoS1 | cameo54321: 1929 is about base classes etc, then the actual impleentations need the boost/stan bindings, but we can do this then one -by-one | 18:38 |
HeikoS1 | cameo54321: I gotta go now. But feel free to play with all that stuff, thats appreciated ! | 18:39 |
cameo54321 | Ok. Thanks. | 18:39 |
-!- HeikoS1 [~heiko@p4FED43A8.dip0.t-ipconnect.de] has quit [Ping timeout: 265 seconds] | 18:43 | |
-!- kislay [~kislay@93.114.45.13] has joined #shogun | 18:47 | |
-!- kislay [~kislay@93.114.45.13] has quit [Ping timeout: 244 seconds] | 18:59 | |
-!- sunil1337 [0e8bd057@gateway/web/freenode/ip.14.139.208.87] has joined #shogun | 19:06 | |
-!- ckkay [~ckkay@chello080108038237.1.11.vie.surfer.at] has quit [Quit: Leaving] | 19:08 | |
@wiking | lisitsyn: magma seems to be really nice: /* Check if matrix is very small then just call LAPACK on CPU, no need for GPU */ | 19:08 |
@lisitsyn | wiking: ha | 19:09 |
@lisitsyn | wiking: what's the API? | 19:09 |
@lisitsyn | e.g. how do you solve a system with magma? | 19:09 |
-!- kislay [~kislay@93.114.45.13] has joined #shogun | 19:11 | |
@wiking | lisitsyn: https://github.com/bravegag/eigen-magma | 19:11 |
@wiking | lisitsyn: and the api is like fucking lapack/blas | 19:11 |
@lisitsyn | wiking: hah | 19:11 |
@wiking | magma_dsyevd_gpu | 19:12 |
@lisitsyn | wiking: so magma is actually just a blas implementation? | 19:12 |
@wiking | ... | 19:12 |
@lisitsyn | hahah | 19:12 |
@lisitsyn | DSYEVD AVADA KEDAVRA MTHFK | 19:12 |
@wiking | yep | 19:12 |
@wiking | indeed | 19:12 |
@lisitsyn | no I actually understand dsyevd and whatever | 19:12 |
@lisitsyn | dsymv :D | 19:12 |
@lisitsyn | dgemm | 19:12 |
@wiking | lisitsyn: https://github.com/bravegag/eigen-magma-benchmark | 19:12 |
@lisitsyn | but indeed it is not very userfriendly | 19:12 |
@lisitsyn | oh? | 19:13 |
@lisitsyn | why | 19:13 |
@lisitsyn | :D | 19:13 |
@lisitsyn | wiking: eigen <-> magma bridge is a very bad idea | 19:14 |
@wiking | why what? | 19:14 |
@lisitsyn | wiking: why it has that cool performance | 19:14 |
@wiking | :DDD | 19:14 |
@wiking | not always | 19:14 |
@lisitsyn | wiking: the more data you put the more gflops you get | 19:14 |
@lisitsyn | will this ever end :D | 19:14 |
@wiking | as u can see sometimes it's not worth because cpu<->gpu mem transfer takes too long | 19:14 |
@lisitsyn | yeah sure | 19:14 |
@wiking | but yeah magma seems to be cool | 19:15 |
@wiking | now the question is of course | 19:15 |
@lisitsyn | I want to play with that but I still don't have any gpu haha | 19:15 |
@wiking | how are we fucking going to interface this :))) | 19:15 |
@wiking | lisitsyn: 06:04.0 VGA compatible controller: Matrox Graphics, Inc. MGA G200eW WPCM450 (rev 0a) | 19:15 |
@wiking | :DDDD | 19:15 |
@wiking | that's in fatbot | 19:15 |
@lisitsyn | wiking: matrox? is that something from 95? | 19:15 |
@lisitsyn | 1995 I mean | 19:15 |
@wiking | yep | 19:16 |
@lisitsyn | 3dfx? | 19:16 |
@lisitsyn | :D | 19:16 |
@lisitsyn | voodoo graphics? | 19:16 |
@wiking | hahahahaha | 19:17 |
@wiking | yeah remember those :) | 19:17 |
@wiking | lisitsyn: this is cool https://github.com/bravegag/eigen-magma/blob/master/Eigen/src/Eigenvalues/SelfAdjointEigenSolver_MAGMA.h | 19:20 |
@lisitsyn | wiking: yeah though well it could be done w/o macroses :D | 19:21 |
@wiking | hahahahha | 19:22 |
@wiking | i was not refering to the actual implementation ;) | 19:22 |
@wiking | jsut the idea :) | 19:22 |
@besser82 | lisitsyn, wiking: thou shalt use pocl with magma to to get thy openCL-backend ;) | 19:24 |
@besser82 | lisitsyn, wiking: will work even without gpu | 19:25 |
@wiking | besser82: u mean clMagma? | 19:25 |
@besser82 | wiking, yes, magma has support for different backend afaik?!? | 19:26 |
@besser82 | wiking, which should be interchangeable.... | 19:26 |
@wiking | besser82: yeah magma has like 3 differnt implementation. magma, clMagma and magma mic | 19:27 |
@wiking | that Intel Xeon Phi Coprocessors seems like heavyweight shit | 19:27 |
@wiking | it's like FPU of the 2000s ;P | 19:28 |
@besser82 | wiking: but they are more cool than fpu :D | 19:29 |
-!- sunil1337 [0e8bd057@gateway/web/freenode/ip.14.139.208.87] has quit [Ping timeout: 245 seconds] | 19:29 | |
@wiking | because everything is cooler in 2000s | 19:29 |
@wiking | ;) | 19:29 |
@besser82 | wiking, not everything | 19:32 |
@besser82 | wiking, nuthing beats C64 with 2MBytes RAM :P | 19:33 |
-!- sunil1337 [0e8bd057@gateway/web/freenode/ip.14.139.208.87] has joined #shogun | 19:35 | |
-!- Saurabh7 [~Saurabh7@115.248.130.148] has quit [Quit: HydraIRC -> http://www.hydrairc.com <- Go on, try it!] | 19:43 | |
-!- sunil1337 [0e8bd057@gateway/web/freenode/ip.14.139.208.87] has quit [Ping timeout: 245 seconds] | 19:52 | |
-!- sonne|osx [~sonne@e178249159.adsl.alicedsl.de] has joined #shogun | 19:56 | |
-!- thoralf [~thoralf@p4FEC8826.dip0.t-ipconnect.de] has joined #shogun | 19:59 | |
thoralf | Hey! | 19:59 |
-!- cameo54321 [27293c04@gateway/web/freenode/ip.39.41.60.4] has quit [Ping timeout: 245 seconds] | 20:01 | |
-!- kislay [~kislay@93.114.45.13] has quit [Read error: Connection reset by peer] | 20:16 | |
-!- buravec [~buravec@46.148.31.104] has joined #shogun | 20:22 | |
-!- kislay [~kislay@93.114.45.13] has joined #shogun | 20:32 | |
-!- buravec [~buravec@46.148.31.104] has quit [Read error: Connection reset by peer] | 20:41 | |
-!- kislay [~kislay@93.114.45.13] has quit [Read error: Connection reset by peer] | 20:44 | |
-!- kislay [~kislay@93.114.45.13] has joined #shogun | 20:58 | |
-!- kislay [~kislay@93.114.45.13] has quit [Quit: Leaving] | 21:12 | |
-!- kislay [~kislay@93.114.45.13] has joined #shogun | 21:12 | |
-!- sunil1337 [0e8bd057@gateway/web/freenode/ip.14.139.208.87] has joined #shogun | 21:18 | |
-!- abinash_panda [01261735@gateway/web/cgi-irc/kiwiirc.com/ip.1.38.23.53] has joined #shogun | 21:28 | |
-!- sonne|osx [~sonne@e178249159.adsl.alicedsl.de] has quit [Quit: sonne|osx] | 21:30 | |
-!- sonne|osx [~sonne@e178249159.adsl.alicedsl.de] has joined #shogun | 21:36 | |
abinash_panda | Hi all !! I am Abinash | 21:39 |
abinash_panda | I am new to shogun. I am interested in working on the issue https://github.com/shogun-toolbox/shogun/issues/1929 Unify Shogun's probability distribution. If anybody can help me in this regard. | 21:39 |
-!- abinash_panda [01261735@gateway/web/cgi-irc/kiwiirc.com/ip.1.38.23.53] has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client] | 21:44 | |
-!- abinash_panda [~abinash@1.38.23.53] has joined #shogun | 21:45 | |
thoralf | Hey Abinash! | 21:45 |
thoralf | abinash: Did you subscribe to the shogun-list yet? | 21:45 |
abinash_panda | lisitsyn, wiking sonney2k : If you could help in understanding the issue https://github.com/shogun-toolbox/shogun/issues/1929 Unify Shogun's probability distribution | 21:46 |
abinash_panda | thoralf, Yes I have subscribed to the shogun-list | 21:46 |
thoralf | abinash: Heiko wrote this issue, so maybe ask him on the list. | 21:46 |
abinash_panda | thoralf: Ok. | 21:47 |
abinash_panda | When can I find him on IRC or should I ask him over the mailing list ? | 21:48 |
thoralf | abinash: The point is, that we have implementations of needed distributions in different places and that we want to unify them. I think HeikoS (his IRC name) got inspired by STAN. | 21:48 |
thoralf | He's on IRC many times a day. If you're lucky, he'll be in today, but better write to the list. | 21:49 |
abinash_panda | thoralf: Thanks. I would wait for him and I would also write on the list. | 21:51 |
thoralf | abinash_panda: You might want to check the logs for his last visit: http://bit.ly/1eBecWl | 21:55 |
-!- thoralf [~thoralf@p4FEC8826.dip0.t-ipconnect.de] has quit [Quit: Konversation terminated!] | 22:12 | |
-!- sunil1337 [0e8bd057@gateway/web/freenode/ip.14.139.208.87] has quit [Quit: Page closed] | 22:38 | |
-!- lambday [67157f54@gateway/web/freenode/ip.103.21.127.84] has joined #shogun | 23:13 | |
-!- lambday [67157f54@gateway/web/freenode/ip.103.21.127.84] has quit [Ping timeout: 245 seconds] | 23:33 | |
--- Log closed Sat Mar 08 00:00:29 2014 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!