--- Log opened Fri Jul 13 00:00:17 2012 | ||
heiko | blackburn, how do you mean that? | 00:18 |
---|---|---|
heiko | git revision stuff? | 00:18 |
blackburn | heiko: well we need to have git revision in footer or so | 00:19 |
heiko | ah alright | 00:19 |
heiko | mmmmmmh | 00:19 |
heiko | no idea actually, but I could check that out :) | 00:19 |
heiko | blackburn, http://thorehusfeldt.net/2011/05/13/including-git-revision-identifiers-in-latex/ | 00:21 |
blackburn | yeah I didn't like that solution | 00:21 |
blackburn | but if you don't know - no need to do that | 00:21 |
heiko | yeh this one is too complicated ;) | 00:22 |
heiko | we could also do a trick | 00:22 |
heiko | just load it from a file or so | 00:22 |
heiko | with a package to include text from files | 00:22 |
heiko | or even \input | 00:22 |
heiko | and a maklefile that generates the .tex file with the revision | 00:22 |
blackburn | yeah probably makefile makes sense here | 00:24 |
@sonney2k | blackburn, what did they do in eigen3? | 00:24 |
blackburn | sonney2k: template lazy evaluation, etc | 00:24 |
blackburn | a + b + c makes no copies | 00:25 |
@sonney2k | ok | 00:25 |
@sonney2k | blackburn, sounds very expensive though | 00:25 |
blackburn | sonney2k: expensive at? | 00:26 |
@sonney2k | compiling | 00:26 |
blackburn | yes, a little | 00:26 |
CIA-18 | shogun: Sergey Lisitsyn openmp * r1d1648f / (3 files in 2 dirs): Make use of openmp locks instead of pthread ones - http://git.io/fqHkLA | 00:26 |
blackburn | sonney2k: but code is much more easier to digest and faster | 00:26 |
@sonney2k | sure | 00:26 |
@sonney2k | and if we only use this in .cpp files then it is no problem | 00:27 |
CIA-18 | shogun: Sergey Lisitsyn openmp * r773b253 / src/shogun/kernel/Kernel.h : Made Kernel use openmp instead of pthread - http://git.io/2NIxbg | 00:27 |
@sonney2k | heiko, cool that you start doxygen latex stuff too :) | 00:27 |
@sonney2k | blackburn, do you know what gsomix wanted to ask? | 00:28 |
blackburn | nope | 00:28 |
heiko | sonney2k, yeah I like that idea :) I would have loved if a student did this book thing over the summer, but well, now we just do it ourselves | 00:28 |
heiko | sonney2k, blackburn, currently searching for inspiration on structuring things ... any ideas? I might get a copy of the weka book, but their slides dont look nice ... | 00:29 |
CIA-18 | shogun: Sergey Lisitsyn openmp * ra3ee63e / (3 files in 2 dirs): Removed unnecessary pthread includes - http://git.io/ykggRw | 00:29 |
blackburn | good catch I'm in search for inspiration | 00:29 |
@sonney2k | well I don't know - either we do it as ML reference and algorithm/method based | 00:30 |
@sonney2k | or as a problem based one | 00:30 |
@sonney2k | e.g. have text -> use string kernels with svms or whatever | 00:30 |
heiko | I wanted to start by describing the methods, so algorithm based | 00:30 |
@sonney2k | vs. SVM -> kernel -> lala | 00:30 |
heiko | like what part of the api does what | 00:30 |
heiko | and then maybe in addition a problem based thing | 00:31 |
@sonney2k | I am also a big fan of the algorithm based stuff | 00:31 |
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has quit [Ping timeout: 252 seconds] | 00:31 | |
heiko | then its like a reference | 00:31 |
@sonney2k | but in this case we can describe things in separate sections | 00:31 |
blackburn | we can combine both | 00:31 |
@sonney2k | and structure it later | 00:31 |
@sonney2k | blackburn, how? | 00:31 |
heiko | people dont know what shogun can do, I just talked to a post-doc here who wasnt aware of all the things it can do | 00:31 |
blackburn | sonney2k: first algorithms then applications | 00:31 |
heiko | homepage should be simplified too so that one gets a quick overview | 00:32 |
@sonney2k | true one can give some applications | 00:32 |
blackburn | heiko: can you help me with openmping? | 00:32 |
@sonney2k | n4nd0, speaking of which - where is your br4ther ? | 00:32 |
heiko | blakcburn, maybe, what are the exactly doing? | 00:32 |
blackburn | heiko: you would need to unparallelize parallelized thing and then just set up omp pragma | 00:33 |
@sonney2k | blackburn, how is it going with openmp? how many algos did you transition? | 00:33 |
heiko | sonney2k, explaining things, I wouldnt include that in there. If we do the 100th SVM explanation, nobody cares, and theres loads of good literature | 00:34 |
blackburn | sonney2k: going | 00:34 |
blackburn | sonney2k: if you want to help - join ;) | 00:34 |
heiko | sonney2k, but describing the algorithms, like what do they do, when do they work, numerical stuff, parameters, and all the stuff thats not covered in normal textbooks | 00:34 |
heiko | blackburn, do you got an example? | 00:34 |
@sonney2k | heiko, well you have to briefly describe the method nonetheless | 00:35 |
blackburn | heiko: https://github.com/shogun-toolbox/shogun/commit/4e109e36c992a423171dcd98bd6cb9aa8f971bd6 | 00:35 |
@sonney2k | blackburn, I am not an openmp expert | 00:36 |
heiko | blackburn, is that already the parallelizing? | 00:36 |
blackburn | heiko: yes | 00:36 |
heiko | because so much more deletions than additions | 00:36 |
heiko | wow | 00:36 |
blackburn | heiko: all one need to get for parallel is to write down private variables | 00:36 |
blackburn | and may be schedule and nowait options | 00:37 |
heiko | blackburn, what about thread safety and stuff | 00:37 |
heiko | nothing in shogun is safe | 00:37 |
blackburn | heiko: I can't see any difference | 00:38 |
blackburn | heiko: what we got with pthreads is not safer | 00:38 |
heiko | currently, one has ultimate controll over parallelized things, if there is just an easy api, one might oversee things perhaps? | 00:38 |
@sonney2k | heiko, for 99.999% of the stuff you don't need thread safety | 00:38 |
heiko | mmh, well ok | 00:39 |
@sonney2k | e.g. traversing in r/o way some buffer | 00:39 |
heiko | would be cool for grid-searching and x-val | 00:39 |
@sonney2k | wiking_, long time no updates... could you please say a word? | 00:39 |
heiko | but actually, we would have to change so many things | 00:39 |
@sonney2k | blackburn, please tell gsomix that he should rather ask directly not ask if I am there... I usually try to read the logs | 00:40 |
blackburn | heiko: no, it is a misbelief you control anything in pthreads | 00:40 |
blackburn | you just do useless things, not control | 00:40 |
@sonney2k | PROPAGANDA :D | 00:40 |
heiko | sonney2k, blackburn, I got another thing: the model-selection framework only works on CMachine, but now I got these statistics classes .... | 00:40 |
heiko | lol blackburn, :) | 00:40 |
heiko | blackburn, I agree, also pthreads is ugly | 00:41 |
heiko | and openmp looks nice :D | 00:41 |
blackburn | compiler can set up params->start and params->end by itself | 00:41 |
blackburn | it is not actually control | 00:41 |
@sonney2k | heiko, then create some CStatisticsMachine out of it | 00:41 |
@sonney2k | err them | 00:41 |
heiko | sonney2k, yeah I had that in mind, but things become a bit more complicated | 00:42 |
heiko | because the evaluation also has to be abstracted more | 00:42 |
@sonney2k | blackburn, do you have some openmp handbook? | 00:42 |
heiko | you optimize over statistics | 00:42 |
heiko | anyway, Ill first finish the other stuff I want to do | 00:42 |
heiko | sonney2k, wikipedia has some examples | 00:43 |
blackburn | sonney2k: http://www.openmp.org/mp-documents/OpenMP3.0-SummarySpec.pdf | 00:43 |
n4nd0 | sonney2k: hey! I am talking to bernard right now on gmail, do you want me to tell him to come around? | 00:45 |
blackburn | https://computing.llnl.gov/tutorials/openMP/ sonney2k | 00:45 |
@sonney2k | n4nd0, I just want to know if he continues work ... | 00:46 |
n4nd0 | sonney2k: asking... | 00:46 |
heiko | sonney2, btw are all of the mid-term evals in? | 00:46 |
@sonney2k | yes | 00:47 |
heiko | nice | 00:47 |
blackburn | sonney2k: heiko: btw you may help me in totally different way | 00:47 |
heiko | blackburn, yes? | 00:47 |
blackburn | just write unit-tests :D | 00:47 |
blackburn | cause I do not test it right now and it should be thoroughly tested before merging | 00:48 |
heiko | argh | 00:48 |
heiko | please write a test for everything that is changed | 00:48 |
heiko | before you change it | 00:48 |
blackburn | it is not fun | 00:48 |
heiko | all my migration code currently dont work due to changes | 00:48 |
n4nd0 | sonney2k: he's working on the news issue, he says he'll appear tomorrow to talk to you | 00:48 |
heiko | and one has review all the code all over again | 00:48 |
blackburn | heiko: this is all impossible with such heavy changes | 00:51 |
heiko | blackburn, I agree, but I think its very dangerous changing things too fast without knowing whether something is broken. And when change A broke something but you notice after changes B,C,D, this is so hard to figure out | 00:52 |
blackburn | too late | 00:52 |
blackburn | :D | 00:52 |
blackburn | well one thing we could try is to rewrite everything | 00:53 |
heiko | crazy russian! ;) | 00:53 |
heiko | sonney2k, blackburn, I gotta go now, have a good night! | 00:54 |
blackburn | heiko: you do underestimate the power of vodka | 00:54 |
heiko | blackburn, lol :) well unfortunately I work worse when I drink. I tried it a few times, but it seems to be the wrong way around for me | 00:55 |
@sonney2k | n4nd0, k | 00:55 |
blackburn | heiko: not drunk enough | 00:55 |
blackburn | :D | 00:56 |
@sonney2k | blackburn, there is this comic showing the right balance of alcohol can increase productivity :D | 00:56 |
blackburn | ballmer's peak? | 00:57 |
@sonney2k | https://xkcd.com/323/ | 00:57 |
@sonney2k | exactly | 00:57 |
@sonney2k | heiko, ^ | 00:57 |
@sonney2k | no need to reach that | 00:57 |
heiko | hehe, funny ;) | 00:58 |
@sonney2k | s/no/you | 00:58 |
@sonney2k | going to bed too. | 00:58 |
heiko | good night! | 00:58 |
-!- heiko [~heiko@host86-180-42-236.range86-180.btcentralplus.com] has left #shogun [] | 00:58 | |
-!- naywhayare [~ryan@spoon.lugatgt.org] has quit [Ping timeout: 248 seconds] | 01:11 | |
-!- naywhayare [~ryan@spoon.lugatgt.org] has joined #shogun | 01:12 | |
n4nd0 | blackburn: good night! | 01:16 |
n4nd0 | I think we were the only ones left | 01:16 |
blackburn | heh | 01:16 |
blackburn | good night | 01:16 |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving] | 01:16 | |
-!- blackburn [~blackburn@109.226.92.17] has quit [Quit: Leaving.] | 01:42 | |
shogun-buildbot | build #20 of nightly_none is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_none/builds/20 | 03:00 |
-!- Netsplit *.net <-> *.split quits: wiking_ | 03:56 | |
-!- Netsplit over, joins: wiking_ | 04:15 | |
-!- romi_ [~mizobe@187.11.98.72] has quit [Remote host closed the connection] | 06:10 | |
-!- pluskid [~pluskid@li411-226.members.linode.com] has joined #shogun | 07:19 | |
-!- gsomix [~gsomix@r206-10.smr.ru] has joined #shogun | 08:09 | |
gsomix | good morning | 08:09 |
-!- zxtx [~zv@67.159.191.98] has joined #shogun | 08:31 | |
@sonney2k | gsomix, morning! | 08:50 |
@sonney2k | gsomix, what did you want to ask yesterday? | 08:50 |
@sonney2k | (but didn't...) | 08:50 |
gsomix | sonney2k, any suggestion about buffer stuff | 08:52 |
@sonney2k | gsomix, I liked what I have seen | 08:56 |
@sonney2k | gsomix, I just don't know if it works with python2.x too | 08:56 |
@sonney2k | (that is the only reason why I did not yet merge) | 08:57 |
-!- uricamic [~uricamic@2001:718:2:1634:5506:9b36:907d:3c5] has joined #shogun | 08:57 | |
gsomix | sonney2k, hm, I use python2.7 for buffer protocol coding | 08:57 |
gsomix | sonney2k, but don't merge it now. | 08:58 |
gsomix | I will send PR at evening with buffer for other DenseFeatures classes | 08:58 |
gsomix | (now I'm at supervisor. doing some optics stuff) | 08:59 |
@sonney2k | gsomix, ok this evening then :) | 09:00 |
gsomix | btw my optics work looks like gamedev... ray tracing | 09:07 |
@sonney2k | gsomix, chances are not high that you convert me to optics :D | 09:14 |
@sonney2k | blackburn had quite some strong (negative) influence about this subject *lol* | 09:14 |
gsomix | haha | 09:17 |
@sonney2k | but I guess it is all exaggerated as usual :) | 09:18 |
-!- zxtx [~zv@67.159.191.98] has quit [Ping timeout: 240 seconds] | 09:20 | |
CIA-18 | shogun: Soeren Sonnenburg master * r20bc943 / (2 files in 2 dirs): fix compile errors in slep and mmd when lapack is not available - http://git.io/9tvwVg | 09:31 |
-!- yoo [2eda6d52@gateway/web/freenode/ip.46.218.109.82] has joined #shogun | 09:38 | |
-!- heiko [~heiko@host86-180-42-236.range86-180.btcentralplus.com] has joined #shogun | 09:50 | |
CIA-18 | shogun: Heiko Strathmann master * rc464e7a / src/shogun/statistics/LinearTimeMMD.cpp : fixes a compile error when LAPACK isnt installed - http://git.io/H0uNsg | 10:04 |
CIA-18 | shogun: Heiko Strathmann master * r4d4855e / src/shogun/statistics/LinearTimeMMD.cpp : Merge pull request #645 from karlnapf/master - http://git.io/7CcVeA | 10:04 |
-!- gsomix [~gsomix@r206-10.smr.ru] has quit [Remote host closed the connection] | 10:05 | |
-!- yoo [2eda6d52@gateway/web/freenode/ip.46.218.109.82] has quit [Quit: Page closed] | 10:17 | |
-!- yoo [2eda6d52@gateway/web/freenode/ip.46.218.109.82] has joined #shogun | 10:27 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 10:32 | |
-!- emrecelikten is now known as emre-awaj | 10:59 | |
-!- emre-awaj is now known as emre-away | 10:59 | |
-!- heiko [~heiko@host86-180-42-236.range86-180.btcentralplus.com] has quit [Ping timeout: 265 seconds] | 10:59 | |
-!- bern4rd [4f940120@gateway/web/freenode/ip.79.148.1.32] has joined #shogun | 11:02 | |
bern4rd | hi | 11:04 |
n4nd0 | what's up bern4rd | 11:17 |
n4nd0 | sonney2k: around? | 11:17 |
bern4rd | ey n4nd0 | 11:18 |
@sonney2k | n4nd0, yes? | 11:20 |
n4nd0 | sonney2k: I have a quick doubt about the hm-svm, I think you know about it | 11:22 |
n4nd0 | sonney2k: the reason why the PLiFs are used is because the observations are real data, isn't it? | 11:22 |
CIA-18 | shogun: Soeren Sonnenburg master * rfdd7763 / src/shogun/statistics/LinearTimeMMD.cpp : Revert "fixes a compile error when LAPACK isnt installed" - http://git.io/SrtYLw | 11:23 |
@sonney2k | well they mostly are yes | 11:23 |
@sonney2k | but plifs enable some richer spaces... | 11:23 |
@sonney2k | bern4rd, so whats up with the website? are you still working on it? | 11:24 |
n4nd0 | sonney2k: richer spaces for what? | 11:24 |
@sonney2k | pluskid, could you email me some sho - gun kanjis that fit? | 11:24 |
bern4rd | yes, I worked on the news | 11:24 |
@sonney2k | n4nd0, instead of just having a single real valued observation | 11:24 |
@sonney2k | you can have a piece-wise linear function | 11:25 |
pluskid | sonney2k: do you want to line up horizontally? | 11:25 |
@sonney2k | pluskid, you are the better designer ... do it as you think is best | 11:25 |
pluskid | sonney2k: I'm just thinking that the logo area for the new site needs horizontal picture | 11:26 |
pluskid | sonney2k: I'll give you some examples | 11:26 |
n4nd0 | sonney2k: ok, thank you. I think it is very interesting that one can deal with real data, I'd never thought of it before | 11:26 |
@sonney2k | n4nd0, so you can approximate basically any function (if you had unlimited helper points) | 11:26 |
@sonney2k | n4nd0, I guess I don't understand - what kind of data did you usually work with? | 11:27 |
n4nd0 | sonney2k: aham I see | 11:27 |
@sonney2k | pluskid, ok | 11:27 |
@sonney2k | bern4rd, should we move the shogun-web git over to official shogun-toolbox? | 11:27 |
n4nd0 | sonney2k: always that I had used HMMs the observations where inside a finite space | 11:27 |
bern4rd | the structure of the news is going to be always the same? because there is a little difference between news from 2011/2012 and older news. | 11:28 |
@sonney2k | bern4rd, yes | 11:28 |
@sonney2k | n4nd0, even for HMMs there are continuos output variants (with e.g. gaussian distributed outputs) | 11:28 |
n4nd0 | sonney2k: aham | 11:29 |
bern4rd | as you want sonney2k, i need to work more on it | 11:29 |
n4nd0 | brb | 11:30 |
@sonney2k | bern4rd, I might have some time in future evenings (not so clear but maybe :) so when I am bored I could also work on things | 11:30 |
bern4rd | =) ok as you prefer then | 11:32 |
bern4rd | where in shogun git should I upload the files of the web app? | 11:37 |
bern4rd | sonney2k: ? | 11:48 |
-!- bern4rd [4f940120@gateway/web/freenode/ip.79.148.1.32] has quit [Quit: Page closed] | 12:09 | |
-!- pluskid [~pluskid@li411-226.members.linode.com] has quit [Quit: Leaving] | 12:22 | |
@sonney2k | bern4rd - I will create a new one shogun-web | 12:26 |
@sonney2k | pluskid - I like #3 | 12:26 |
-!- pluskid [~pluskid@li411-226.members.linode.com] has joined #shogun | 12:27 | |
@sonney2k | pluskid - I like #3 & #8 | 12:32 |
@sonney2k | but I just sent you an email... | 12:32 |
pluskid | sonney2k: so I guess the agreement is on #3 :) | 12:47 |
pluskid | and we need to and english sho and gun below the kan ji | 12:47 |
shogun-buildbot | build #21 of nightly_none is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_none/builds/21 | 12:54 |
-!- pluskid [~pluskid@li411-226.members.linode.com] has quit [Ping timeout: 255 seconds] | 13:04 | |
-!- alexlovesdata [~binder@goldenezahl.ml.tu-berlin.de] has joined #shogun | 13:20 | |
-!- pluskid [~pluskid@111.120.83.239] has joined #shogun | 13:22 | |
-!- heiko [~heiko@host86-183-73-88.range86-183.btcentralplus.com] has joined #shogun | 13:52 | |
-!- heiko [~heiko@host86-183-73-88.range86-183.btcentralplus.com] has quit [Read error: Connection reset by peer] | 13:55 | |
-!- heiko [~heiko@host86-183-73-88.range86-183.btcentralplus.com] has joined #shogun | 13:57 | |
-!- emre-away is now known as emrecelikten | 14:02 | |
-!- pluskid [~pluskid@111.120.83.239] has quit [Ping timeout: 255 seconds] | 14:10 | |
-!- pluskid [~pluskid@li411-226.members.linode.com] has joined #shogun | 14:11 | |
yoo | why do we have a such assertion ASSERT(str->have_same_length()) when compute Weighted Degreee Kernel feature space explicitly ?? | 14:14 |
n4nd0 | yoo: weigthed degree string kernel? | 14:25 |
yoo | no CWDFeatures for example | 14:27 |
yoo | it explicitly compute the new features right ? | 14:27 |
n4nd0 | I think that not necessarily | 14:28 |
yoo | instead of evaluating the kernel, I would like first to see the features in the kernel space explicitly. | 14:28 |
n4nd0 | yoo: there must be a way to do it, but I have never tried so | 14:29 |
yoo | allright then | 14:30 |
yoo | thx | 14:30 |
n4nd0 | yoo: take a look to the methods of CKernel, I'd say that there must be something to help you doing that | 14:31 |
-!- pluskid [~pluskid@li411-226.members.linode.com] has quit [Quit: Leaving] | 14:40 | |
-!- heiko [~heiko@host86-183-73-88.range86-183.btcentralplus.com] has quit [Quit: Leaving.] | 14:51 | |
-!- heiko [~heiko@host86-183-73-88.range86-183.btcentralplus.com] has joined #shogun | 14:59 | |
n4nd0 | someone using valgrind around? | 15:00 |
n4nd0 | there's a thing that I don't quite understand and it kind of bugs me | 15:00 |
-!- heiko [~heiko@host86-183-73-88.range86-183.btcentralplus.com] has quit [Client Quit] | 15:01 | |
uricamic | n4nd0: yep, working with it right now | 15:40 |
n4nd0 | uricamic: I found a nice description and I taking a look to it, http://es.gnu.org/~aleksander/valgrind/valgrind-memcheck.pdf | 15:41 |
n4nd0 | but thank you! | 15:41 |
uricamic | no problem :) | 15:41 |
uricamic | btw if you have 64bit system you might have some problems with valgrind | 15:41 |
uricamic | to be more precise if you are using some precompiled package | 15:42 |
n4nd0 | I use 32bit linux so the problems shouldn't arise, I guess | 15:42 |
uricamic | probably not | 15:43 |
alexlovesdata | what is tyour problem with valgrind nanno? | 15:43 |
uricamic | on 64bit there is some problem with some instructions which are not supported by valgrind yet | 15:43 |
alexlovesdata | does compile with -O2 help? | 15:43 |
uricamic | and also you need to compile it from source code | 15:43 |
alexlovesdata | or you just do not see the precise lines? | 15:44 |
n4nd0 | alexlovesdata: I didn't how to interpret correctly the leak summary | 15:47 |
alexlovesdata | because you do not see the precise line for some commands right? | 15:47 |
alexlovesdata | then try compiling with -g3 -fno-inline | 15:47 |
alexlovesdata | some commands get strange (eg C++ std containers) when you do not use -fno-inline | 15:48 |
n4nd0 | alexlovesdata: aham, ok, thank you very much! | 15:48 |
n4nd0 | I have just realized that probably I was losing lot of info because I was not using the flag -g when compiling | 15:48 |
alexlovesdata | -g3 is even more verbose | 15:48 |
n4nd0 | btw, shouldn't definitely lost be always bigger than indirectly lost? | 15:50 |
n4nd0 | I understand definitely groups directly and indirectly lost, that's why | 15:50 |
n4nd0 | I didn't say anything | 15:51 |
n4nd0 | definitely lost = directly lost in valgrind's summary | 15:51 |
n4nd0 | leak summary, to be precise | 15:52 |
-!- zxtx [~zv@64.25.27.130] has joined #shogun | 15:57 | |
alexlovesdata | http://valgrind.org/docs/manual/faq.html#faq.deflost | 16:18 |
alexlovesdata | explains it I think :) | 16:18 |
-!- alexlovesdata [~binder@goldenezahl.ml.tu-berlin.de] has left #shogun [] | 16:28 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving] | 16:30 | |
-!- zxtx [~zv@64.25.27.130] has quit [Ping timeout: 240 seconds] | 16:42 | |
-!- uricamic [~uricamic@2001:718:2:1634:5506:9b36:907d:3c5] has quit [Quit: Leaving.] | 17:09 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 17:44 | |
-!- gsomix [~gsomix@83.149.21.187] has joined #shogun | 18:04 | |
-!- puffin444 [62e3926e@gateway/web/freenode/ip.98.227.146.110] has joined #shogun | 18:12 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Ping timeout: 248 seconds] | 18:33 | |
-!- bern4rd [4f940120@gateway/web/freenode/ip.79.148.1.32] has joined #shogun | 18:41 | |
-!- gsomix [~gsomix@83.149.21.187] has quit [Ping timeout: 244 seconds] | 18:55 | |
-!- puffin444 [62e3926e@gateway/web/freenode/ip.98.227.146.110] has quit [Ping timeout: 245 seconds] | 19:02 | |
-!- gsomix [~gsomix@85.26.233.175] has joined #shogun | 19:09 | |
-!- bern4rd [4f940120@gateway/web/freenode/ip.79.148.1.32] has quit [Quit: Page closed] | 19:17 | |
-!- alexlovesdata [~binder@goldenezahl.ml.tu-berlin.de] has joined #shogun | 19:21 | |
-!- bern4rd [4f940120@gateway/web/freenode/ip.79.148.1.32] has joined #shogun | 19:26 | |
-!- yoo [2eda6d52@gateway/web/freenode/ip.46.218.109.82] has quit [Ping timeout: 245 seconds] | 19:28 | |
-!- gsomix [~gsomix@85.26.233.175] has quit [Ping timeout: 255 seconds] | 19:46 | |
-!- puffin444 [62e3926e@gateway/web/freenode/ip.98.227.146.110] has joined #shogun | 20:00 | |
-!- blackburn [~blackburn@109.226.92.17] has joined #shogun | 20:09 | |
@sonney2k | bern4rd, looks pretty cool already ! | 20:10 |
bern4rd | thanks :) | 20:11 |
@sonney2k | bern4rd, pluskid will send us some kanji for the logo but it is pretty neat already | 20:11 |
bern4rd | ok nice, there are a lot of pictures that should be modified in the home view | 20:12 |
@sonney2k | yeah | 20:12 |
@sonney2k | I guess we can project some machine learning method's output in there | 20:12 |
@sonney2k | so it looks like real science ;-) | 20:12 |
bern4rd | yes! | 20:13 |
bern4rd | ah I forget, where I should upload the webapp in shogun git? | 20:14 |
blackburn | sonney2k: what is going on with logo? ;) | 20:19 |
blackburn | can I see variants? | 20:19 |
blackburn | and site pic may be too | 20:20 |
alexlovesdata | I wanna see the logo, too. hehe | 20:22 |
-!- gsomix [~gsomix@85.26.233.175] has joined #shogun | 20:23 | |
blackburn | everybody wants to see the logo, sonney2k! | 20:24 |
alexlovesdata | I agree with blackburn | 20:26 |
-!- heiko [~heiko@host86-183-73-88.range86-183.btcentralplus.com] has joined #shogun | 20:31 | |
-!- heiko [~heiko@host86-183-73-88.range86-183.btcentralplus.com] has quit [Client Quit] | 20:31 | |
-!- bern4rd [4f940120@gateway/web/freenode/ip.79.148.1.32] has quit [Quit: Page closed] | 20:32 | |
-!- alexlovesdata [~binder@goldenezahl.ml.tu-berlin.de] has left #shogun [] | 21:14 | |
-!- gsomix [~gsomix@85.26.233.175] has quit [Ping timeout: 246 seconds] | 21:30 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 21:41 | |
blackburn | n4nd0: hey | 21:43 |
-!- gsomix [~gsomix@85.26.235.226] has joined #shogun | 21:43 | |
n4nd0 | hey blackburn | 21:44 |
n4nd0 | I am going to have dinner now | 21:44 |
n4nd0 | tty later | 21:44 |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Client Quit] | 21:44 | |
blackburn | argh | 21:44 |
blackburn | n4nd0: wanted to ask you about mosek :D | 21:44 |
gsomix | sonney2k, be my PR manager :) | 21:44 |
@sonney2k | gsomix, looks nice | 22:01 |
@sonney2k | gsomix, did you test if it works / compiles? | 22:01 |
gsomix | sonney2k, yes, I did... but one moment. last test | 22:02 |
@sonney2k | gsomix, what I don't understand is why you did add an extra buffer_helpers.i | 22:06 |
@sonney2k | and also why you include it in modshogun.i but not just in features.i | 22:06 |
gsomix | >> add an extra buffer_helpers.i | easier to find | 22:08 |
gsomix | >> modshogun.i but not just in features.i | hm, my fault, yes | 22:08 |
gsomix | sonney2k, hm, not fault. buffer_helpers.i will contain all helper functions for all buffered classes. | 22:11 |
gsomix | I think, that include "buffer_helpers.i" should be global at modshogun.i | 22:11 |
@sonney2k | gsomix, ok then please rename it to python_buffer_helpers.i or so | 22:11 |
@sonney2k | I see | 22:12 |
@sonney2k | it is python only so python_ + what is the official name? | 22:12 |
@sonney2k | buffer_protocol? | 22:12 |
@sonney2k | python_buffer_protocol.i then right? | 22:13 |
gsomix | yep | 22:13 |
blackburn | gsomix: sonney2k so what can we do now? | 22:13 |
blackburn | features[:,1] = ...? | 22:13 |
gsomix | blackburn, I did not add sliceing (sequence protocol). | 22:15 |
blackburn | gsomix: so features[i,1] = 1.0? | 22:15 |
@sonney2k | gsomix, did you do the rename? | 22:16 |
gsomix | blackburn, nope. it's sequence protocol, not buffer. at now we can do something like DenseFeatures + numpy.arrays and numpy.arrays(DenseFeatures) | 22:16 |
gsomix | sonney2k, yes, I did | 22:18 |
gsomix | sonney2k, btw I should to write nice example | 22:20 |
@sonney2k | gsomix, then pleae update the PR | 22:20 |
gsomix | ok | 22:21 |
@sonney2k | such that I can merge | 22:21 |
@sonney2k | gsomix, I think next steps are slicing and support also for operators such as += *= /= -= etc to make things like x[0,0]=17 and x+=y possible | 22:23 |
@sonney2k | gsomix, next time also please use git commit --amend to not turn this into an extra commit | 22:23 |
-!- romi_ [~mizobe@187.11.98.72] has joined #shogun | 22:23 | |
CIA-18 | shogun: Evgeniy Andreev master * r40b1246 / (2 files): buffer protocol for DenseFeatures - http://git.io/0ZGAdA | 22:23 |
CIA-18 | shogun: Evgeniy Andreev master * ra184880 / (2 files in 2 dirs): preliminary design of buffer protocol stuff - http://git.io/3kv5Wg | 22:23 |
CIA-18 | shogun: Soeren Sonnenburg master * ra6d49a5 / (4 files in 3 dirs): Merge pull request #642 from gsomix/buffer_protocol - http://git.io/NB_5GQ | 22:23 |
CIA-18 | shogun: Evgeniy Andreev master * r08b1c6c / src/interfaces/python_modular/buffer_helpers.i : buffer stuff separated to file - http://git.io/PdZvWg | 22:23 |
CIA-18 | shogun: Evgeniy Andreev master * rcaae3ab / (3 files in 2 dirs): renamed buffer helper module - http://git.io/AGgO1Q | 22:23 |
gsomix | sonney2k, sorry, forgot =___= | 22:24 |
gsomix | sonney2k, can you test buffer stuff now? | 22:24 |
@sonney2k | gsomix, btw for operator stuff it should suffice to overload the operators | 22:31 |
@sonney2k | C++ operators I mean | 22:31 |
gsomix | sonney2k, and also for slicing? | 22:32 |
gsomix | ok, I'll try | 22:32 |
blackburn | sonney2k: did you get mosek license? I remember you wrote to smb from mosek | 22:32 |
@sonney2k | gsomix, for slicing this cannot work... slicing can do more e.g. x[0:10,0] etc | 22:33 |
@sonney2k | gsomix, modshogun_wrap.cxx:333104:25: warning: deprecated conversion from string constant to ‘char*’ [-Wwrite-strings] | 22:33 |
@sonney2k | modshogun_wrap.cxx: In function ‘int CharFeatures_getbuffer(PyObject*, Py_buffer*, int)’: | 22:33 |
gsomix | yes | 22:34 |
@sonney2k | ^I get lots of warnings like this | 22:34 |
gsomix | I'll fix it little later. | 22:34 |
@sonney2k | so maybe you just add the (char*) cast | 22:34 |
gsomix | it's stange that in python3 view->format is const char* and in python2.7 view->format is char* | 22:36 |
shogun-buildbot | build #137 of deb3 - modular_interfaces is complete: Failure [failed compile ruby_modular] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/137 blamelist: Evgeniy Andreev <gsomix@gmail.com> | 22:36 |
@sonney2k | gsomix, they fixed it for 3.0 it seems | 22:37 |
gsomix | sonney2k, shogun-buildbot oh no... | 22:37 |
@sonney2k | gsomix, seems to work - nice job \o/ | 22:38 |
@sonney2k | gsomix, I guess you had some intermediate broken commit | 22:38 |
@sonney2k | and forgot about using --amend to fix it. | 22:38 |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 22:38 | |
n4nd0 | blackburn: hey, sorry about before | 22:38 |
n4nd0 | blackburn: tell me, what's up with mosek? | 22:38 |
@sonney2k | blackburn, I didn't follow up on this | 22:38 |
gsomix | sonney2k,not broken commit | 22:38 |
blackburn | n4nd0: did you ever try to combine it with octave? | 22:39 |
n4nd0 | blackburn: no, although the hmsvm toolbox I am using supports it | 22:39 |
n4nd0 | I use it from matlab though | 22:39 |
gsomix | sonney2k, buffer stuff added in global modular interface | 22:39 |
blackburn | argh don't want to install matlab :D | 22:39 |
blackburn | sonney2k: did you? | 22:40 |
gsomix | sonney2k, how can I check what's modular interface compiling? | 22:40 |
@sonney2k | I don't have it | 22:40 |
n4nd0 | blackburn: what do you need to know? | 22:40 |
@sonney2k | gsomix, ? | 22:40 |
n4nd0 | blackburn: we can check it in the hmsvm toolbox probably | 22:40 |
@sonney2k | gsomix, look at waterall | 22:40 |
@sonney2k | it says which is compiling | 22:40 |
blackburn | n4nd0: does hmsvm make use of mosek? | 22:40 |
@sonney2k | but I guess any non python interface will fail | 22:41 |
n4nd0 | blackburn: yes, mosek or cplex | 22:41 |
-!- puffin444 [62e3926e@gateway/web/freenode/ip.98.227.146.110] has quit [Ping timeout: 245 seconds] | 22:41 | |
n4nd0 | blackburn: no standalone | 22:41 |
gsomix | sonney2k, fail in modulars | 22:41 |
gsomix | sonney2k, look in Features.i. I use #ifdef HAVE_PYTHON | 22:41 |
gsomix | BUFFER_DENSEFEATURES(CharFeatures, char, char, "c\0") | 22:41 |
gsomix | #endif | 22:41 |
blackburn | hmm I think I need to add it to path | 22:42 |
gsomix | sonney2k, is it right? | 22:42 |
@sonney2k | gsomix, look at revision a184880c06fae3c283a6bc24cbe79d6481b98ad4 | 22:42 |
@sonney2k | gsomix, you didn't have the HAVE_PYTHON there | 22:42 |
@sonney2k | and forgot to use commit --amend later | 22:42 |
@sonney2k | gsomix, once it builds your latest commit it will likely go back to green | 22:43 |
n4nd0 | blackburn: from hmsvm toolbox's readme http://pastebin.com/0dHc1Lwj | 22:44 |
n4nd0 | it might help | 22:44 |
blackburn | n4nd0: so they use some wrapper for that | 22:45 |
-!- puffin444 [62e3926e@gateway/web/freenode/ip.98.227.146.110] has joined #shogun | 22:45 | |
n4nd0 | blackburn: yeah, there are some files here | 22:45 |
n4nd0 | blackburn: do you want to take a look at them? | 22:45 |
n4nd0 | http://mloss.org/software/tags/hmsvm/ | 22:45 |
blackburn | yeah could you please send they? | 22:45 |
blackburn | aham | 22:45 |
n4nd0 | I think it is faster if you just get it there | 22:45 |
n4nd0 | probably src/opt_interface/mosek is interesting for you | 22:46 |
shogun-buildbot | build #138 of deb3 - modular_interfaces is complete: Failure [failed compile ruby_modular] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/138 blamelist: Evgeniy Andreev <gsomix@gmail.com>, Soeren Sonnenburg <sonne@debian.org> | 22:49 |
CIA-18 | shogun: Sergey Lisitsyn openmp * rf517565 / (2 files): OpenMP in KernelMachine - http://git.io/4RSAVQ | 23:03 |
gsomix | sonney2k, can I see you tomorrow online? | 23:08 |
-!- heiko [~heiko@host86-183-73-88.range86-183.btcentralplus.com] has joined #shogun | 23:11 | |
-!- heiko [~heiko@host86-183-73-88.range86-183.btcentralplus.com] has quit [Quit: Leaving.] | 23:27 | |
gsomix | http://cs309420.userapi.com/v309420232/3133/vc5-tBt_hcM.jpg | 23:35 |
gsomix | too small :( | 23:36 |
n4nd0 | wow I live in IKEA :D | 23:36 |
gsomix | http://cs309420.userapi.com/v309420232/3135/2Ztsii0fbW4.jpg | 23:36 |
emrecelikten | Workforce source :D | 23:36 |
gsomix | http://cs309420.userapi.com/v309420232/313e/1_lNqfNbJqE.jpg | 23:36 |
blackburn | n4nd0: group sex lol :D | 23:36 |
n4nd0 | haha | 23:36 |
n4nd0 | I have missed that!! | 23:37 |
n4nd0 | miss like that I have not found it around here | 23:37 |
n4nd0 | saying that I missed was ambiguous :DD | 23:37 |
blackburn | :D | 23:37 |
blackburn | rofl | 23:37 |
emrecelikten | n4nd0: Hahah | 23:37 |
emrecelikten | Perhaps you "missed" the feeling too. "Ah it's been a while" | 23:38 |
n4nd0 | yeah .. that's why | 23:38 |
n4nd0 | I realized, fuck, what have I just said!?! | 23:38 |
n4nd0 | haha | 23:38 |
blackburn | heiko lives right in abramovich | 23:39 |
n4nd0 | hehe | 23:39 |
emrecelikten | http://farm6.static.flickr.com/5086/5371734858_f4f4664057_o.jpg | 23:42 |
emrecelikten | Such a simple world view, not surprised | 23:43 |
blackburn | nordturkei :D | 23:47 |
n4nd0 | http://www.elia.ws/blog/wp-content/uploads/2011/11/europe-according-to-spain.jpg | 23:51 |
n4nd0 | blackburn: married priests? | 23:51 |
blackburn | verry strange :D | 23:52 |
n4nd0 | haha Iceland = fridge | 23:52 |
gsomix | all maps btw http://alphadesigner.com/mapping-stereotypes/ | 23:54 |
blackburn | that's a good shot http://alphadesigner.com/wp-content/uploads/world-according-to-israel.jpg | 23:56 |
n4nd0 | hehe | 23:58 |
--- Log closed Sat Jul 14 00:00:17 2012 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!