--- Log opened Thu Aug 23 00:00:17 2012 | ||
shogun-buildbot_ | build #63 of nightly_none is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_none/builds/63 | 03:13 |
---|---|---|
shogun-buildbot_ | build #74 of nightly_default is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_default/builds/74 | 03:44 |
-!- emrecelikten1 [~emre@trir-4d0d8caa.pool.mediaWays.net] has joined #shogun | 04:26 | |
-!- emrecelikten [~emre@trir-5d8000e4.pool.mediaWays.net] has quit [Ping timeout: 248 seconds] | 04:27 | |
-!- av3ngr [av3ngr@nat/redhat/x-kzxmpmjwtrcqadom] has joined #shogun | 05:56 | |
-!- av3ngr [av3ngr@nat/redhat/x-kzxmpmjwtrcqadom] has quit [Client Quit] | 05:56 | |
-!- cheng [80fa500f@gateway/web/freenode/ip.128.250.80.15] has joined #shogun | 08:16 | |
-!- uricamic [~uricamic@2001:718:2:1634:6dc1:2e18:bbfe:a408] has joined #shogun | 08:35 | |
-!- blackgun|shoburn [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has joined #shogun | 08:37 | |
blackgun|shoburn | hey cheng | 08:37 |
CIA-52 | shogun: Michal Uricar master * r3713c0c / examples/undocumented/libshogun/so_multiclass_BMRM.cpp : BMRM libshogun example fix (runnable without cmd arguments) - http://git.io/8yy6bA | 08:40 |
CIA-52 | shogun: Michal Uricar master * rf1a5b95 / examples/undocumented/python_modular/graphical/so_multiclass_BMRM.py : so_multiclass_BMRM.py fix - http://git.io/51HZpQ | 08:40 |
CIA-52 | shogun: Sergey Lisitsyn master * rdea19d8 / (2 files in 2 dirs): Merge pull request #754 from uricamic/BM_SOL_EXAMPLE - http://git.io/jn1uiQ | 08:40 |
cheng | blackburn: Hi | 08:43 |
cheng | blackgun|shoburn: Hi again. :-) | 08:43 |
blackgun|shoburn | okay let get not so crazy nickname | 08:44 |
-!- blackgun|shoburn is now known as blackburn | 08:44 | |
cheng | blackburn: You want to start a new channel so that we don't flood shogun? | 08:45 |
blackburn | we could just send messages directly | 08:45 |
blackburn | check other tab | 08:45 |
shogun-buildbot_ | build #337 of bsd1 - libshogun is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/337 | 08:52 |
shogun-buildbot_ | build #448 of deb3 - modular_interfaces is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/448 | 09:18 |
shogun-buildbot_ | build #355 of deb2 - static_interfaces is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/355 | 09:24 |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 10:22 | |
-!- emrecelikten1 [~emre@trir-4d0d8caa.pool.mediaWays.net] has quit [Quit: Leaving.] | 11:15 | |
-!- blackburn [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has quit [Quit: Page closed] | 11:16 | |
-!- cheng [80fa500f@gateway/web/freenode/ip.128.250.80.15] has quit [Ping timeout: 245 seconds] | 11:23 | |
wiking | yo | 12:03 |
wiking | n4nd0: ping | 12:06 |
n4nd0 | wiking: tell me | 12:14 |
wiking | when u have time i want to discuss a patch with you | 12:14 |
n4nd0 | I have time now | 12:15 |
wiking | it's a massively ugly patch | 12:15 |
wiking | but i really dont know what to do about this | 12:15 |
n4nd0 | mine? | 12:15 |
wiking | nono mine | 12:15 |
n4nd0 | ok | 12:15 |
wiking | mmm i think i'll just pastebin it | 12:15 |
wiking | n4nd0: header http://pastebin.com/a8EUQEu7 | 12:16 |
wiking | implementation http://pastebin.com/pBcjZ7LB | 12:17 |
wiking | this is the consequence of not being able to have class anything: public CStructuredModel, public CLatentModel | 12:17 |
n4nd0 | yeah, I see | 12:19 |
n4nd0 | but it could be bad to have multiple inheritance ... | 12:19 |
n4nd0 | why are there two classes, CLSOModel and CLatentSOModel? | 12:19 |
wiking | well the naming is now stupid | 12:20 |
n4nd0 | aham I see | 12:20 |
wiking | but basically CLSOModel goes for the structural machine | 12:20 |
n4nd0 | one extends LatenModel and the other one the StructuredModel | 12:20 |
wiking | yes | 12:20 |
wiking | indeed | 12:20 |
wiking | and in order to just | 12:21 |
wiking | require one inheritance | 12:21 |
wiking | i hardwired CLatentSOModel into CLSOModel | 12:21 |
wiking | i.e. a user will only need to do an inharitance of CLatentSOModel | 12:21 |
wiking | and define all the relevant application specific functions | 12:21 |
n4nd0 | I understand | 12:22 |
n4nd0 | I guess that, providing how it is one right now, a user should do inheritance of CLSOModel or? | 12:22 |
n4nd0 | I mean it since it is the class that has access to CLatentSOModel and not vice versa | 12:23 |
wiking | no | 12:23 |
wiking | it should do inheritance of CLatentSOModel | 12:23 |
wiking | the CLSOModel is completely fixed | 12:23 |
wiking | and basically it's 'just' a proxy class | 12:23 |
n4nd0 | ok | 12:24 |
wiking | and we cannot do multiple inheritance as then some modular interfaces will fail | 12:25 |
wiking | like java | 12:25 |
wiking | so we either go with something like this | 12:25 |
wiking | or i just throw out latentmodel as is | 12:25 |
wiking | and try to use structuredmodel | 12:25 |
wiking | the problem is that for simple LatentSVM | 12:25 |
wiking | most of the stuff in structuredmodel is not necessary | 12:26 |
wiking | and some other stuff are missing... | 12:26 |
wiking | so if possible i would like to keep it | 12:26 |
wiking | and maybe just rename CLSOModel to CLatentSOProxy | 12:26 |
wiking | ;D | 12:26 |
n4nd0 | that last renaming is good I think | 12:27 |
n4nd0 | :) | 12:27 |
n4nd0 | but I don't think it is such an ugly patch! | 12:28 |
n4nd0 | I mean it is not a common thing to do | 12:28 |
n4nd0 | but it is not a common situation either | 12:28 |
n4nd0 | I think it is ok | 12:28 |
wiking | btw: this is lol: http://gizmodo.com/5937143/what-facebook-deals-with-everyday-27-billion-likes-300-million-photos-uploaded-and-500-terabytes-of-data | 12:28 |
wiking | 500 Terabytes of Data /day | 12:28 |
n4nd0 | dafuq | 12:29 |
wiking | yeah | 12:29 |
wiking | and it's all running on hadoop | 12:29 |
wiking | pretty impressive that it can handle it | 12:29 |
wiking | hahaha the hdfs is 100+ PB | 12:30 |
wiking | :DDD | 12:30 |
n4nd0 | smart people | 12:30 |
wiking | yeah maybe shogun would fail on trying to use this :) | 12:30 |
wiking | StreamingFeatures ;) | 12:30 |
n4nd0 | hehe | 12:31 |
wiking | ok well if u don't find this patch soooo ugly | 12:32 |
wiking | then i can actually commit | 12:32 |
wiking | just need to finish calculating the subgrad in the risk function | 12:32 |
n4nd0 | ok | 12:33 |
wiking | do u actually use in PrimalMosek the y value in the ResultSet? | 12:35 |
wiking | or just psi_truth and psi_pred | 12:36 |
wiking | ? | 12:36 |
n4nd0 | wiking: you mean the argmax right? | 12:36 |
wiking | yeah sorry | 12:37 |
wiking | yeah ResultSet->argmax | 12:37 |
n4nd0 | I think not | 12:37 |
n4nd0 | let me check | 12:37 |
wiking | yeah i've greped now | 12:37 |
wiking | didn't show up | 12:37 |
wiking | just wanted to make sure | 12:37 |
n4nd0 | no | 12:37 |
n4nd0 | just delta and psis | 12:37 |
n4nd0 | that one is used in the moment of prediction | 12:38 |
n4nd0 | in the PrimalMosek we are training so ignored | 12:38 |
wiking | k | 12:38 |
wiking | thnx a lot | 12:38 |
n4nd0 | you are welcome | 12:39 |
n4nd0 | this is just awesome | 12:42 |
n4nd0 | http://www.heraldo.es/noticias/cultura/2012/08/21/la_restauracion_una_pintura_mural_una_iglesia_borja_acaba_chapuza_200865_308.html | 12:42 |
n4nd0 | I don't know if you understand | 12:42 |
n4nd0 | but just look at the picture... | 12:42 |
n4nd0 | restoration, they said | 12:44 |
wiking | yeah i saw that... the old lady went a little bit creative ;) | 13:07 |
wiking | uricamic: here? | 13:16 |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Ping timeout: 276 seconds] | 13:16 | |
uricamic | wiking: yep | 13:16 |
wiking | so i'm just checking the multiclassmodel again, the risk function | 13:16 |
wiking | and the subgrad is R'({\bf w}) = \sum_{i=1}^{m} \Psi(x_i, \hat{y}_i) - \Psi(x_i, y_i), | 13:16 |
wiking | right? | 13:16 |
uricamic | yep | 13:17 |
wiking | i wanted to say that couldn't we just use the CResultSet->psi_truth and CResultSet->psi_pred | 13:17 |
wiking | but then i've realised that argmax is not being called in the ::risk function | 13:18 |
uricamic | not yet, but eventually it should be called there | 13:18 |
wiking | ok ok | 13:18 |
wiking | then i guess we could use those two values | 13:18 |
wiking | as they are generated already by argmax | 13:18 |
uricamic | but the psi_pred has to be computed including the loss | 13:18 |
uricamic | I mean argmax, computed on features infected by loss | 13:19 |
wiking | ok i think this one is ready let's test it on the example :) | 13:32 |
-!- cheng [73406f11@gateway/web/freenode/ip.115.64.111.17] has joined #shogun | 13:57 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 14:03 | |
-!- cheng [73406f11@gateway/web/freenode/ip.115.64.111.17] has quit [Ping timeout: 245 seconds] | 14:22 | |
wiking | AAAAAAAAAAAAAAAAAAAA | 14:28 |
n4nd0 | wiking: ? | 14:33 |
wiking | screaming at my code :) | 14:33 |
wiking | why no work :) | 14:33 |
n4nd0 | does it work? | 14:33 |
n4nd0 | I mean when you scream :D | 14:33 |
wiking | heheh yeah | 14:35 |
@sonney2k | wiking, please do your final evaluations today! | 15:04 |
wiking | hahah i looove when somebody is creating a release with -g flags | 15:04 |
@sonney2k | ? | 15:04 |
wiking | you always get some nice info: | 15:04 |
wiking | /home/eda/mosekprj/brn-6-0/extern/flexlm/src/11.10.0/osx64x86/libsb.a | 15:04 |
@sonney2k | why would one not? | 15:04 |
wiking | sonney2k: in a release? | 15:04 |
@sonney2k | ahh closed source ok... | 15:04 |
wiking | sonney2k: even if it's open source | 15:05 |
@sonney2k | ? | 15:05 |
wiking | i wouldn't create a release with -g | 15:05 |
@sonney2k | then you have no info when things crash | 15:05 |
@sonney2k | wiking, seen the 'please do the final eval today'? | 15:05 |
wiking | sonney2k: affirmitive | 15:05 |
wiking | sonney2k: *nod* and on it! | 15:05 |
wiking | sonney2k: i mean honestly i would just do a striped library for real release | 15:06 |
wiking | that way small memory footprint | 15:06 |
wiking | etc | 15:06 |
wiking | but it's good now as my code crashed big time | 15:06 |
wiking | ah null pointer referencing :) | 15:07 |
wiking | juhuuu | 15:07 |
wiking | n4nd0: still here? | 15:07 |
n4nd0 | wiking: yeah | 15:07 |
n4nd0 | uricamic: are you around? | 15:07 |
wiking | have u checked the static analyzer report for structure/* | 15:07 |
uricamic | n4nd0: yep | 15:07 |
n4nd0 | wiking: not really to be honest | 15:08 |
wiking | n4nd0: http://maeth.com/shogun-report/report-LoMVrV.html#EndPath | 15:08 |
wiking | do u use DynProg atm somewhere? | 15:08 |
wiking | i mean it's jsut that it'd be great to clear out some of those problems | 15:08 |
n4nd0 | uricamic: have you analyzed the Dual Lib QP BM SOSVM for memory leaks? I get some in my example when I am using it :S | 15:09 |
@sonney2k | wiking, how do you create the static analysis btw? | 15:09 |
wiking | i'm happy to create a new report with the latest commit | 15:09 |
@sonney2k | wiking, I mean we could run this on the bb's regularly... | 15:09 |
n4nd0 | wiking: I am using PLiFs but those are defined in Plif.h and so not in DynProg | 15:09 |
wiking | sonney2k: yeah but we need clang/llvm for it | 15:09 |
wiking | sonney2k: at least for this particular one | 15:09 |
n4nd0 | wiking: so no, I am not using anything from DynProg directly | 15:09 |
uricamic | n4nd0: to be honest not | 15:09 |
uricamic | I mean not using e.g. valgrind | 15:09 |
wiking | sonney2k: clang-analyzer.llvm.org | 15:09 |
@sonney2k | wiking, that shouldn't be the problem I think | 15:09 |
uricamic | but according to some report system it should be ok | 15:10 |
n4nd0 | uricamic: aham ok, I think there may be some leaks there - at least it is what valgrind tells me | 15:10 |
wiking | sonney2k: i'm happy to do it if u give me access where i can have clang | 15:10 |
uricamic | n4nd0: ok, can u send me the log then? | 15:10 |
n4nd0 | uricamic: if you are busy with another thing I can try to fix it if you don't mind | 15:10 |
uricamic | n4nd0: ok, feel free to do that :) | 15:11 |
wiking | sonney2k: http://clang-analyzer.llvm.org/installation.html | 15:11 |
n4nd0 | uricamic: I will let you know about any change | 15:11 |
wiking | sonney2k: i'm not 100% sure that u really need to build the whole thing from scratch | 15:11 |
uricamic | n4nd0: ok | 15:11 |
uricamic | thx | 15:11 |
wiking | sonney2k: but yeah it'd be cool to have that report within the buildbot system | 15:12 |
wiking | hahahaha | 15:12 |
wiking | sonney2k: http://packages.ubuntu.com/search?searchon=contents&keywords=scan-build&mode=filename&suite=precise&arch=any | 15:12 |
wiking | sonney2k: easy peasy ;) | 15:12 |
wiking | sonney2k: apt-get install clang | 15:13 |
wiking | sonney2k: after that you just need scan-build ./configure && scan-build make | 15:14 |
wiking | and your report is ready | 15:14 |
@sonney2k | wiking, will that do the usual compile too? | 15:14 |
@sonney2k | I mean is the analysis in addition? | 15:15 |
wiking | sonney2k: good question | 15:19 |
wiking | sonney2k: it did compile me the libshogun | 15:20 |
n4nd0 | wiking: have you used resize_vector in SGVector? | 15:22 |
n4nd0 | wiking: valgrind detects leaks with it but I don't see anything weird in the code | 15:22 |
wiking | n4nd0: noup i haven't | 15:24 |
-!- blackburn [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has joined #shogun | 15:33 | |
blackburn | worst day on job ever :D | 15:37 |
n4nd0 | blackburn: why so? | 15:40 |
blackburn | support is on fire with 3-4 blocker issue | 15:40 |
blackburn | may be even 5 | 15:40 |
n4nd0 | what is that? | 15:42 |
blackburn | issue - well just like github issue | 15:42 |
blackburn | blocker means we have to respond immediately or close to that | 15:42 |
n4nd0 | aaahm | 15:44 |
n4nd0 | it was the blocker what confused me hehe | 15:44 |
n4nd0 | but that means some job to do, right? | 15:44 |
@sonney2k | blackburn, so down to 0 blockers now? | 15:45 |
@sonney2k | blackburn, btw any ideas where gsomix is hiding? | 15:45 |
blackburn | n4nd0: blocker is like OMG DAMN WE ARE GOING TO DIE issue | 15:46 |
blackburn | sonney2k: no I am still wroking on it | 15:46 |
n4nd0 | blackburn: hehe | 15:46 |
n4nd0 | blackburn: I have a blocker issue here now | 15:46 |
@sonney2k | blackburn, I at least know what you are talking about | 15:46 |
blackburn | sonney2k: hehe | 15:46 |
n4nd0 | blackburn: valgrind complains quite a bit of SGVector::resize_vector, any clue about it? | 15:46 |
n4nd0 | have you used this method before? | 15:46 |
blackburn | n4nd0: no, not really | 15:47 |
blackburn | let me get back to that a little bit later | 15:47 |
blackburn | :) | 15:47 |
n4nd0 | I'll investigate a bit more then | 15:47 |
@sonney2k | n4nd0, it is like hey we sold you an alarm clock but it will stop incrementing seconds after 7 days of use | 15:47 |
n4nd0 | sonney2k: :D I like your metaphors | 15:48 |
n4nd0 | but last day's one with the party to go by car or boat and the bundle methods for SO is remains my favourite :D | 15:49 |
n4nd0 | uricamic: let me ask you, why do you use resize_vector in libbmrm.cpp lines 528-530 | 15:51 |
n4nd0 | ? | 15:51 |
blackburn | sonney2k: I hope you didn't have to use clearcase | 15:51 |
uricamic | n4nd0: it is there just because of the history tracks of Fp, Fd and wdists | 15:52 |
n4nd0 | uricamic: it looks to me that at that point there is nothing else to do in the algorithm, then why to resize those vectors? | 15:52 |
blackburn | and we, wiking and I, are the only victims of it | 15:52 |
uricamic | and resize is there since I guess in the beginning there is initialization to the BufSize | 15:52 |
uricamic | so to remove the rest after nIter | 15:53 |
n4nd0 | uricamic: aham, I understand | 15:53 |
uricamic | but probably this could be done in more clever way now | 15:53 |
uricamic | I was writing it in a hurry and also changed the way how these values are represented, so it is possible that the recent version is cumbersome | 15:54 |
n4nd0 | uricamic: no problem :) | 15:55 |
uricamic | so the lines valgrind was complaining about were just these with resize_vector ? | 15:56 |
n4nd0 | uricamic: yes, it seems something is not working as it should with resize_vector | 15:56 |
uricamic | n4nd0: I see, well positive is at least that the rest is ok :) | 15:57 |
n4nd0 | uricamic: there was another thing related to using get_label in MulticlassModel::risk | 15:57 |
n4nd0 | uricamic: but that was easy to solve | 15:57 |
uricamic | n4nd0: I see, ok | 15:58 |
n4nd0 | uricamic: when you call get_label you need to do an UNREF later since this method increases the reference count of the object that returns | 15:58 |
n4nd0 | uricamic: I will show you when I prepare the patch | 15:58 |
uricamic | n4nd0: aha, ok, thx | 15:58 |
@sonney2k | n4nd0, btw is your friend bern4rd still alive? I hope I can install the new website rather soon that is why I am asking (in case we run into issues...) | 16:00 |
n4nd0 | sonney2k: yes he is alive, he's in an internship in Paris | 16:01 |
n4nd0 | sonney2k: I can tell him about it | 16:01 |
@sonney2k | Cool thanks | 16:02 |
n4nd0 | what day will that be? | 16:04 |
@sonney2k | n4nd0, well we put it live September eleven errr 1 :) | 16:05 |
@sonney2k | if blackburn doesn't have blockers to fix :D | 16:05 |
@sonney2k | but I want to put it live under some 'hidden' url fist | 16:05 |
@sonney2k | first | 16:05 |
blackburn | next 3 days should be free | 16:05 |
n4nd0 | sonney2k: ok | 16:05 |
blackburn | that's why I am finishing things with my burning head | 16:05 |
@sonney2k | blackburner! | 16:06 |
@sonney2k | bunsenblackburner even! | 16:06 |
blackburn | lol | 16:07 |
n4nd0 | uricamic: https://github.com/shogun-toolbox/shogun/pull/756/files | 16:20 |
uricamic | n4nd0: THANKS | 16:21 |
n4nd0 | that has solved the leaks I saw using so_multiclass.cpp | 16:21 |
n4nd0 | it seems that there are some others triggered when using so_multiclass_BMRM.cpp | 16:21 |
uricamic | ok, I will have to update valgrind and look on that | 16:22 |
n4nd0 | uricamic: I posted in my blog a good reference for valgrind if you need to check something | 16:23 |
-!- yoo [575b08cb@gateway/web/freenode/ip.87.91.8.203] has joined #shogun | 16:23 | |
yoo | hi all | 16:23 |
n4nd0 | hi yoo | 16:24 |
uricamic | n4nd0: the only problem I encountered was that the last version I had haven't supported some instructions and therefore it actually didn't provide any useful information, the way how that could be overcome was to compile with some flag to disable those features | 16:25 |
n4nd0 | uricamic: aham ok | 16:25 |
uricamic | but somebody told me, that there should be now new version for ubuntu, that should fix that issue | 16:25 |
uricamic | so I hope it will be ok, when I will update it | 16:25 |
n4nd0 | uricamic: but that thing was for python support or for C++ programs too? | 16:26 |
uricamic | C++ | 16:26 |
uricamic | it was something with 64bit and the intel i7, I think | 16:26 |
blackburn | fails on i5 as well | 16:27 |
uricamic | that valgrind was not able to get some instructuons and then it just refuse to examine the code | 16:27 |
blackburn | when I debug things I disable optimizations and it works | 16:27 |
CIA-52 | shogun: thereisnoknife master * r800c845 / src/shogun/CMakeLists.txt : update src/shogun/CMakeLists.txt - http://git.io/mxTAIA | 16:27 |
CIA-52 | shogun: thereisnoknife master * r764509e / src/shogun/CMakeLists.txt : update src/shogun/CMakeLists.txt - http://git.io/VXas-w | 16:27 |
CIA-52 | shogun: Sergey Lisitsyn master * r00d3056 / src/shogun/CMakeLists.txt : Merge pull request #755 from thereisnoknife/master - http://git.io/LoljPg | 16:27 |
uricamic | blackburn: yep, something like that | 16:27 |
uricamic | but the latest release of valgrind should hopefully fix that | 16:28 |
n4nd0 | blackburn: CMakeLists, does shogun support cmake? didn't know about that! | 16:36 |
blackburn | yeah there was a effort of some guy before | 16:37 |
n4nd0 | nice | 16:37 |
yoo | n4nd0: you cannot compile shogun using cmake because the lib dependencies are not up-to-date | 16:46 |
yoo | n4nd0: I have fill the CMakeLists.txt with new files just for visualization | 16:46 |
yoo | n4nd0: my IDE read cmake files, that's why .. | 16:47 |
n4nd0 | yoo: aham, but we will be able to do so, compile with cmake, in the future? | 16:47 |
yoo | n4nd0: then we have to had all the dependencies in the root cmakelists.txt | 16:48 |
yoo | I can try to do this in the future but I am not very used to all shogun dependencies since I am still a shogun-noob | 16:50 |
-!- blackburn [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has quit [Quit: Page closed] | 17:03 | |
-!- yoo [575b08cb@gateway/web/freenode/ip.87.91.8.203] has quit [Quit: Page closed] | 17:06 | |
n4nd0 | sonney2k: around? | 17:15 |
n4nd0 | see you later guys | 17:20 |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has left #shogun [] | 17:20 | |
-!- uricamic [~uricamic@2001:718:2:1634:6dc1:2e18:bbfe:a408] has quit [Quit: Leaving.] | 17:25 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 19:01 | |
n4nd0 | nice, we have users from Ernst & Young now :) | 19:05 |
@sonney2k | n4nd0, ? | 19:11 |
n4nd0 | sonney2k: the guy in the mailing list | 19:21 |
@sonney2k | n4nd0, ahh I meant to reply to your 'around?' | 19:26 |
n4nd0 | sonney2k: ah ok, that was in case you could merge the PR I issued before since I wanted to put some other changes | 19:26 |
n4nd0 | but at the end I just put everything in the same PR | 19:26 |
n4nd0 | it is a small patch in any case though | 19:26 |
CIA-52 | shogun: Soeren Sonnenburg master * r10341ca / (2 files in 2 dirs): Minor warning fix. - http://git.io/CaeoZA | 19:29 |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Ping timeout: 272 seconds] | 19:49 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 20:47 | |
-!- blackburn [~blackburn@188.168.5.81] has joined #shogun | 20:59 | |
blackburn | oh so e&y uses shogun right?/ | 21:00 |
blackburn | :D | 21:00 |
blackburn | interesting news | 21:00 |
n4nd0 | blackburn: yeah :) | 21:01 |
n4nd0 | blackburn: did you see my patch at github? | 21:01 |
blackburn | not yet | 21:01 |
blackburn | what is it about? | 21:01 |
n4nd0 | the generic risk function, I fixed it a bit | 21:01 |
n4nd0 | I get same results with it as with the one michal did in MulticlassModel | 21:02 |
n4nd0 | what is good :) | 21:02 |
blackburn | what was wrong? | 21:02 |
n4nd0 | initialization of subgrad and the computation of the risk | 21:02 |
blackburn | oooops | 21:02 |
n4nd0 | the term -<w, psi(xi,yi)> | 21:03 |
blackburn | do we need to initialize subgrad? | 21:03 |
n4nd0 | yeah | 21:03 |
blackburn | hmm | 21:03 |
blackburn | did you change score in multiclass | 21:03 |
blackburn | I mean now you are adding loss | 21:03 |
blackburn | did you subtract it in multiclass? | 21:03 |
blackburn | see what I mean? | 21:04 |
blackburn | btw I think argmax in training mode should do that | 21:09 |
blackburn | it could be highly inefficient this way | 21:09 |
n4nd0 | highly inefficient because I substract the delta? | 21:13 |
n4nd0 | blackburn: ? | 21:20 |
-!- K0stIa [~kostia@alt2.hk.cvut.cz] has joined #shogun | 22:10 | |
-!- K0stIa [~kostia@alt2.hk.cvut.cz] has left #shogun [] | 22:10 | |
@sonney2k | blackburn, any news on gsomix? | 22:11 |
@sonney2k | blackburn, btw please kick chris that he does your evaluation... | 22:12 |
@sonney2k | and wiking - please do yours too | 22:12 |
blackburn | sonney2k: oops | 22:17 |
blackburn | what is deadline? | 22:17 |
@sonney2k | tomorrow | 22:17 |
blackburn | n4nd0: inefficient because of <w,Phi(x,y_truth)> | 22:17 |
n4nd0 | blackburn: yeah sure, but it is the price of having it generic | 22:18 |
n4nd0 | blackburn: I think the best idea is to do as wiking suggested | 22:18 |
blackburn | we could assume argmax does that | 22:18 |
n4nd0 | blackburn: aham, we could do that too maybe | 22:18 |
blackburn | it sounds more efficient | 22:19 |
n4nd0 | I think so too | 22:20 |
wiking | sonney2k: hye | 22:21 |
wiking | sonney2k: here? | 22:22 |
-!- blackburn1 [~blackburn@188.168.2.190] has joined #shogun | 22:22 | |
wiking | sonney2k: managed to get the scan-build working? | 22:22 |
-!- blackburn [~blackburn@188.168.5.81] has quit [Read error: Operation timed out] | 22:22 | |
blackburn1 | sonney2k: chris says he has done it two days ago | 22:23 |
blackburn1 | sonney2k: are you sure? | 22:24 |
blackburn1 | sonney2k: about gsomix he floats in some space probably - I said to him you are missing him | 22:27 |
-!- gsomix [~gsomix@80.234.22.113] has joined #shogun | 22:28 | |
blackburn1 | sonney2k: hooray | 22:29 |
blackburn1 | n4nd0: lets discuss that issue tomorrow | 22:29 |
n4nd0 | blackburn1: ok | 22:29 |
wiking | blackburn1: yo | 22:29 |
gsomix | sonney2k, sorry for absence. was helping parents with home t last days | 22:29 |
n4nd0 | blackburn1: in any case I think the patch should be ok since the current StructuredModel::risk is wrong | 22:30 |
blackburn1 | wiking: | 22:30 |
wiking | blackburn1: seems EY people will really going to try using shogun for catching tax fraud :DD | 22:30 |
blackburn1 | yeah I've seen | 22:30 |
blackburn1 | sounds crazy but pretty nice to hear | 22:30 |
wiking | it'd be great if he could share us what they are trying to use for | 22:31 |
wiking | i'll email him and ask him... maybe he can share a little bit | 22:31 |
wiking | that reminds me: anybody knows how to prepare an .rpm? | 22:31 |
blackburn1 | hah | 22:31 |
wiking | mmm as i can see the .deb creator scripts are not within the repo either | 22:32 |
wiking | sonney2k do you maybe have them somewhere? | 22:33 |
wiking | i knew once how to create deb package but i think things have changed ever since 2004 :P | 22:34 |
-!- blackburn [~blackburn@188.168.14.160] has joined #shogun | 22:42 | |
-!- blackburn1 [~blackburn@188.168.2.190] has quit [Ping timeout: 248 seconds] | 22:44 | |
wiking | doh: incorrect checksum for freed object | 22:45 |
blackburn | sonney2k: could you please send a mail to chris if there is a problem with my evaluation still | 22:49 |
@sonney2k | blackburn, why don't you do it? | 22:50 |
@sonney2k | gsomix, you have to do your final evaluations! | 22:50 |
@sonney2k | gsomix, ASAP | 22:50 |
blackburn | sonney2k: I said that because I didn't know you are around :) | 22:51 |
blackburn | sonney2k: did you check it just? | 22:51 |
blackburn | is it still not in system? | 22:51 |
@sonney2k | let me check | 22:51 |
gsomix | sonney2k, Final Various usability improvements Submitted 2012-08-23 17:30:03 2012-08-23 17:30:03 | 22:52 |
gsomix | already done, hm | 22:52 |
@sonney2k | alright - everyone submitted | 22:53 |
* sonney2k relaxes | 22:53 | |
blackburn | argh you made all nervous :) | 22:54 |
gsomix | sonney2k, :) dont worry, be happy | 22:54 |
gsomix | ok, it's a working time | 22:54 |
@sonney2k | blackburn, well I just (1 hr ago) received reminder emails - so things have just been done. | 22:56 |
@sonney2k | gsomix, whats the status | 22:56 |
blackburn | as I said chris was sure he did it two days ago | 22:56 |
blackburn | :) | 22:56 |
@sonney2k | he didn't :D | 22:57 |
blackburn | okay anyway it is done and we are all happy | 22:58 |
@sonney2k | blackburn, wait - he did, wiking, gsomix and arthur didn't | 22:58 |
@sonney2k | but now we are back to peace love and you know :D | 22:58 |
gsomix | sonney2k, preparing PR now. plan to talk with heiko tomorrow (luck?) | 22:58 |
@sonney2k | gsomix, you should email him - he is writing his masters thesis currently and probably won't be online | 22:59 |
@sonney2k | (in chat I mean) | 22:59 |
@sonney2k | reachable via email yes | 22:59 |
gsomix | sonney2k, ok, I will do this | 23:00 |
blackburn | see you | 23:06 |
-!- blackburn [~blackburn@188.168.14.160] has quit [Quit: Leaving.] | 23:06 | |
-!- gsomix [~gsomix@80.234.22.113] has quit [Quit: Ex-Chat] | 23:21 | |
n4nd0 | good night guys | 23:24 |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving] | 23:24 | |
--- Log closed Fri Aug 24 00:00:17 2012 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!