Open in new window / Try shogun cloud
--- Log opened Tue Jun 21 00:00:06 2011
CIA-32shogun: Soeren Sonnenburg master * r769d80f / (5 files in 2 dirs): T** transition in wd kernels and sginterface -
CIA-32shogun: Soeren Sonnenburg master * rd20d826 / src/libshogun/classifier/mkl/MKLMultiClass.cpp : fix labels related crasher in mkl multiclass -
@sonney2kblackburn, good luck with your exam tomorrow and good nite01:32
blackburnsonney2k: aren't you sleeping??01:32
* sonney2k falls down the stairs (again!)01:32
blackburnit is on 2201:32
blackburnI have one day for prepare01:32
blackburngood night :)01:32
@sonney2kyes today is 2101:32
blackburnaha, already 21 :)01:33
blackburnit is late even in germany01:33
@sonney2kso prepare01:33
blackburnsonney2k: good night, don't work so much:)01:33
blackburnyeah! got python LMDS01:42
-!- blackburn [~blackburn@] has quit [Quit: Leaving.]02:04
-!- serialhex_ [] has joined #shogun04:02
-!- b3ttyb00 [] has joined #shogun04:05
-!- Netsplit *.net <-> *.split quits: serialhex, @bettyboo04:09
-!- b3ttyb00 is now known as bettyboo04:09
-!- serialhex_ is now known as serialhex05:32
-!- blackburn [~blackburn@] has joined #shogun09:37
-!- heiko1 [] has joined #shogun12:07
-!- f-x [~user@] has joined #shogun12:31
-!- sploving [~root@] has joined #shogun12:46
-!- f-x [~user@] has quit [Ping timeout: 260 seconds]13:21
splovinghello all. git remote add
splovinggit fetch upstream13:34
splovingerror: The requested URL returned error: 40313:34
splovingerror: The requested URL returned error: 40313:34
splovingwhat is the matter?13:35
blackburnwhy do you want to add remote with shogun-toolbox/shogun?13:38
blackburnah I see13:39
blackburnsploving: please paste git remote -v13:39
blackburngit remote rm upstream13:40
blackburngit remote add upstream
splovingthe same error13:41
splovinghave you test it?13:41
blackburnblackburn@blackburn-laptop:~/shogun/shogun/src$ git remote -v13:42 (fetch)13:42 (push)13:42
blackburnupstream (fetch)13:42
blackburnupstream (push)13:42
blackburnblackburn@blackburn-laptop:~/shogun/shogun/src$ git pull upstream master13:43
blackburn * branch            master     -> FETCH_HEAD13:43
blackburnworks ok13:43
splovingI do not know why. my origin git fetch upstream always stops there, like this: Receiving objects:  26% (259/995)13:46
splovingI have to re git clone one13:46
splovinggit clone git clone
splovinggit git remote add upstream
splovingeror, git remote add13:47
splovingbut it shows the following error: error: The requested URL returned error: 40313:47
-!- in3xes [~in3xes@] has quit [Ping timeout: 252 seconds]13:47
-!- in3xes [~in3xes@] has joined #shogun13:48
-!- f-x [~user@] has joined #shogun13:51
CIA-32shogun: Sergey Lisitsyn master * rcbd0931 / src/libshogun/preprocessor/LocallyLinearEmbedding.cpp : ASSERT replacement for LocallyLinearEmbedding -
-!- sploving [~root@] has left #shogun []14:12
-!- heiko1 [] has quit [Ping timeout: 258 seconds]14:14
-!- blackburn [~blackburn@] has quit [Read error: Operation timed out]14:17
-!- blackburn [~blackburn@] has joined #shogun14:18
-!- blackburn [~blackburn@] has quit [Ping timeout: 258 seconds]14:31
-!- blackburn [~blackburn@] has joined #shogun14:32
-!- in3xes_ [~in3xes@] has joined #shogun14:48
-!- in3xes [~in3xes@] has quit [Ping timeout: 240 seconds]14:52
-!- vojta [~quassel@2001:718:2:1634:221:6aff:fe58:151a] has joined #shogun15:44
-!- vojta is now known as VojtechFranc15:44
VojtechFranchi Alesis, are you here?15:45
alesis-novikHello VojtechFranc15:46
VojtechFrancwhat is the progress?15:46
alesis-novikI'm finishing up with the examples, though now I'll need to modify the Gaussian and GMM code to work with Shoguns new SGVector and SGMatrix15:47
alesis-novikI had to do some travelling last week, so I didn't finish the examples. I've done the initial condition optimizations and sampling from the GMM15:47
VojtechFranchave you submit the code to your GIT branch? Can I look at it?15:47
alesis-novikMost of it is there, but I want to finish rewriting other bits before pushing the new things15:48
alesis-novikI feel I'm bad at doing small patches. I keep making large chunks.15:49
VojtechFranchowever, submitting regularly allows to monitor your progrees15:50
VojtechFrancbtw, Soeren remined me that you are not sending your week progress reports.15:51
alesis-novikI missed one because I couldn't send it on Saturday. I wanted to do the next one this Saturday to update on both weeks15:52
VojtechFranchave you submitted any report so far?15:54
alesis-novikI've been doing it every week until this one15:54
VojtechFrancwhere do you send the mails? for some reasons I'm not getting them which, however, might be my fault15:55
alesis-novikI'm sending them to the mailing list. I'll CC them to you from now on as well, if you're not getting any15:57
-!- heiko1 [~heiko@] has joined #shogun15:57
VojtechFrancyes, please do it and also send me please the old ones.15:57
alesis-novikOk, though there won't be anything new in the old ones we haven't talked about.15:58
VojtechFrancyes, still I need to keep track and later use them to write an assesment15:59
alesis-novikSo now that the basic-improved version of EM for GMM is working (I'll finish changing the code and push it tonight), I'd like to chat a bit about component merging and splitting.15:59
alesis-novikI assume that's the next stem16:00
-!- heiko1 [~heiko@] has quit [Client Quit]16:00
VojtechFrancI'd like to play a bit with the basic-improved EM-GMM. I'd like to start with try it the examples you wrote/are writting16:02
VojtechFrancwhen do you expect this part is finished so that I can compile and play16:03
VojtechFrancok, I look forward. Please send me an email when it is ready.16:04
alesis-novikIf somehow my Octave examples won't work, you said I could write Python ones, correct? (I had some trouble setting up octave)16:04
VojtechFrancPython is OK16:05
VojtechFrancactually, I forgot why you can use ordinary Matlab.16:05
alesis-novikI can't use Matlab because I don't have Matlab on the computer I'm programming Shogun on.16:08
VojtechFrancok, I see.16:09
alesis-novikI only use Matlab at the university, and I don't have admin access there16:09
VojtechFrancwhy do you need admin access?16:09
alesis-novikThat and very limited space16:10
alesis-novikWithout admin I'd need to store everything in my allocated space, and I couldn't fit shogun + all dependencies there16:11
VojtechFrancok, no problem16:14
VojtechFrancanyway, I'm looking forward to see tommorow the running examples and please don't forget to send me (all) progress reports. thanks16:15
alesis-novikWill do16:15
-!- in3xes_ is now known as in3xes16:21
VojtechFrancthanks. cheers16:22
-!- VojtechFranc [~quassel@2001:718:2:1634:221:6aff:fe58:151a] has quit [Remote host closed the connection]16:22
blackburnalesis-novik: kinda strong mentoring :)16:42
alesis-novikblackburn, I had a question about the SGVector and SGMatrix16:42
blackburnalesis-novik: shoot16:43
alesis-novikWe are now using them to pass parameters and as returned stuff(?). Should the internal fields be changed to SGVector and SGMatrix as well?16:43
blackburnalesis-novik: not sure everywhere but if it will be consistent16:44
alesis-novikOk. I think it will make sense to use them to sa16:46
blackburnalesis-novik: pros of using SG* is we can encapsulate sizes too16:46
alesis-noviksay store the covariance matrix16:46
alesis-novikblackburn, yeah, it makes things a lot less messy16:46
blackburnyes I guess it would be better - no need to have lots of size variables and so on16:47
@sonney2kspare time!22:44
blackburnsonney2k: how are ya?22:45
@sonney2kblackburn, everyone is asleep now?22:45
blackburnnice :)22:45
blackburnlast exam in 8 hrs!22:46
@sonney2kblackburn, I wish you best of luck!22:47
blackburnsonney2k: yesterday night I wrote some landmark MDS in python, it works just like classic MDS (very similar results) but e.g. took ~30s for 5000 vectors22:47
@sonney2kblackburn, I don't talk to you about shogun things today - you have to learn for your exam.22:48
blackburnsonney2k: not now, have some rest :)22:48
blackburnsonney2k: btw some pull requests awaiting :)22:51
@sonney2kblackburn, you learn for your exam - break is over!22:51
blackburndaway daway rabotay? ;)22:52
serialhexsonney2k: you here??23:00
@sonney2kserialhex, y23:00
serialhexhey, so i figured out why my python wasn't working... it was installing to the wrong dir23:01
serialhexif you go here: it shows you where python stuff is to be installed, and if i had built python from source it would have been fine23:02
@sonney2kserialhex, hmmh ok23:02
serialhexbut b/c i didn't it shows up in /usr/lib/python2.7/site-packages/shogun instead of /usr/local/lib/python2.7/dist-packages/shogun/ (shogun default)23:03
serialhexi'm looking through the src/config file right now for where to change this so it works everywhere, but i'm kind of lost :P (it's a big ass file!!!)23:04
@sonney2kf-x, around?23:04
serialhexi think it's ~lines 1333 or so... but i'm not sure23:05
@sonney2kserialhex, it is weird - they started the dist-packages thing in python2.5 or 2.6 (annoying idea) and everything broke because of that23:05
@sonney2know it is no longer dist-packages? hard to believe23:06
serialhexidk about the dist-packages thing, but it's in /user/lib/python2.7 instead of /user/local/lib/python2.723:08
serialhexyeah, there is no dist-packages dir in my /usr/lib/python2.7 (not user XP )23:09
CIA-32shogun: Shashwat Lal Das master * r096f05b / (145 files in 28 dirs): Merge remote-tracking branch 'upstream/master' into streaming -
CIA-32shogun: Shashwat Lal Das master * r3565447 / (4 files): Made dynamic reallocation possible for features. size for parser is now in terms of ring size wrt number of examples. -
CIA-32shogun: Shashwat Lal Das master * rb85b3fa / (2 files): Switched to vlen nomenclature for SGVector objects instead of length in parser and ring code. -
@sonney2kserialhex, I am afraid the only way to resolve this is to use --pydir as option to configure23:37
CIA-32shogun: Soeren Sonnenburg master * r045dc3a / src/configure : use whitespace in docu -
@sonney2kf-x, please don't forget to rebase instead of merge'ing of upstream/master23:45
@sonney2kf-x, in your patch your code was the master and everything we did was considered the patch now23:45
--- Log closed Wed Jun 22 00:00:10 2011