--- Log opened Thu Oct 13 00:00:39 2011 | ||
-!- blackburn [~blackburn@188.168.4.95] has quit [Quit: Leaving.] | 00:07 | |
-!- alesis-novik [~alesis@dhcp-90-089.inf.ed.ac.uk] has quit [Quit: Leaving] | 01:45 | |
CIA-3 | shogun: Soeren Sonnenburg * r1d5d860 / src/shogun/kernel/Kernel.cpp : Fix a bug in cache_multiple_kernel_row that gets triggered when support vector regression is performed with svr light and multiple threads. - http://git.io/MSI5yQ | 11:12 |
---|---|---|
shogun-buildbot | build #280 of python_static is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/python_static/builds/280 blamelist: sonne@debian.org | 11:19 |
-!- alesis-novik [~alesis@dhcp-90-089.inf.ed.ac.uk] has joined #shogun | 14:01 | |
-!- in3xes [~in3xes@180.149.49.227] has joined #shogun | 18:20 | |
-!- blackburn [~blackburn@188.168.4.89] has joined #shogun | 18:27 | |
-!- in3xes [~in3xes@180.149.49.227] has quit [Ping timeout: 248 seconds] | 18:33 | |
blackburn | uh rip dennis ritchie | 18:34 |
-!- blackburn [~blackburn@188.168.4.89] has quit [Ping timeout: 248 seconds] | 20:20 | |
-!- blackburn [~blackburn@188.168.2.213] has joined #shogun | 20:24 | |
blackburn | sonney2k: I've got an idea of very fun task for next gsoc or just to one of us | 20:33 |
blackburn | there could be some task to store visualization information to make easier to visualize something | 20:34 |
@sonney2k | visualization information? | 20:35 |
blackburn | yeah something related to training process we could plot | 20:36 |
blackburn | sonney2k: I've got a problem with new 'converter' folder, parallel and io are not visible for some reason | 20:37 |
blackburn | any ideas? | 20:37 |
@sonney2k | no | 20:37 |
blackburn | sad | 20:37 |
@sonney2k | blackburn, did you derive from SGObject? | 20:40 |
blackburn | sonney2k: yes, I guess the problem is I forgot to implement get_name | 20:40 |
blackburn | sonney2k: where should I add new modular .i files? | 20:44 |
@sonney2k | in modular/ | 20:44 |
blackburn | sonney2k: no, modshogun.i, somewhere else? | 20:45 |
@sonney2k | ? | 20:47 |
blackburn | nevermind | 20:49 |
blackburn | fixed | 20:59 |
blackburn | sonney2k: where did you declared python modules for workaround like shogun.Preprocessor, shogun.Classifier? | 21:01 |
@sonney2k | blackburn, that is in Makefile.template | 21:01 |
blackburn | thanks | 21:01 |
@sonney2k | blackburn, we have to get memory requirements for compiling the interfaces below 1.5 G | 21:05 |
@sonney2k | better ~1G | 21:05 |
blackburn | oh | 21:05 |
blackburn | why? | 21:05 |
@sonney2k | none of the debs built | 21:05 |
-!- alesis-novik [~alesis@dhcp-90-089.inf.ed.ac.uk] has quit [Quit: Leaving] | 21:27 | |
blackburn | sonney2k: have you asked anyone from swig team how to reduce size? | 21:41 |
@sonney2k | no | 21:43 |
blackburn | may be they have some tricks for that | 21:43 |
@sonney2k | I don't think so... | 21:51 |
-!- sonne|work [~sonnenbu@194.78.35.195] has quit [Ping timeout: 240 seconds] | 22:34 | |
-!- sonne|work [~sonnenbu@194.78.35.195] has joined #shogun | 22:49 | |
blackburn | sonney2k: around? | 23:32 |
CIA-3 | shogun: Soeren Sonnenburg * r74bf297 / src/shogun/kernel/WeightedDegreePositionStringKernel.cpp : fix lhs / rhs mixup in free_feature_vector (how did this ever work?!) - http://git.io/QCICng | 23:35 |
CIA-3 | shogun: Soeren Sonnenburg * ra6e49bb / src/shogun/classifier/svm/WDSVMOcas.h : fix broken free_feature_vector call in wdsvmocas - http://git.io/Q9nfcQ | 23:35 |
CIA-3 | shogun: Soeren Sonnenburg * r12a5173 / (2 files): introduce more efficient cleanup_feature_vectors function - http://git.io/Nx_h0g | 23:35 |
CIA-3 | shogun: Soeren Sonnenburg * r18a93b9 / src/shogun/features/DotFeatures.cpp : always set dot feature property on init - http://git.io/ieu_ow | 23:35 |
CIA-3 | shogun: Soeren Sonnenburg * r6356f26 / (3 files): fix static octave examples - http://git.io/OJKEJQ | 23:35 |
@sonney2k | yes | 23:35 |
@sonney2k | blackburn, busy watching http://www.youtube.com/watch?NR=1&v=pmw87w7i6rQ though | 23:35 |
blackburn | sonney2k: I need an idea how to stole feature matrix from simple features | 23:36 |
blackburn | safely | 23:36 |
@sonney2k | I don't think it is currently possible | 23:41 |
blackburn | sonney2k: I would add leave_feature_matrix(), ok? | 23:43 |
@sonney2k | what exactly should this do? | 23:43 |
@sonney2k | get the ptr of the fm? | 23:43 |
blackburn | sonney2k: yes, and make it not deleted when features will be | 23:43 |
@sonney2k | and then replace the fm ptr with yours? | 23:44 |
blackburn | no | 23:44 |
blackburn | I just want features to lose this pointer | 23:44 |
blackburn | and when it is going to be deleted features won't free this matrix | 23:44 |
blackburn | cause it will be used by another features | 23:45 |
blackburn | I could do this with copying but it is pretty bad way in case of large-scale applications | 23:45 |
@sonney2k | yeah | 23:45 |
@sonney2k | so add a function that gets you the ptr and size | 23:45 |
@sonney2k | and internally drops/resets preprocs/ptr/sizes | 23:46 |
@sonney2k | SGMatrix<ST> get_feature_matrix_and_clear() ? | 23:46 |
@sonney2k | or *_and_reset() ? | 23:46 |
blackburn | why not leave_feature_matrix()? | 23:46 |
@sonney2k | leave doesn't tell anything about that you steal it | 23:47 |
@sonney2k | well how about steal_feature_matrix() | 23:47 |
blackburn | yeah | 23:47 |
blackburn | sonney2k: I have almost done ->converters things | 23:48 |
@sonney2k | blackburn, very good | 23:48 |
blackburn | and already moved all the examples | 23:48 |
blackburn | and tested | 23:48 |
@sonney2k | I think I fixed all octave examples now that octave runs again under debian unstable | 23:48 |
blackburn | final step - dimensionreductionpreprocessors proxy | 23:49 |
shogun-buildbot | build #281 of python_static is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/python_static/builds/281 | 23:49 |
@sonney2k | I have high hopes taht the buildbot goes through now | 23:50 |
@sonney2k | ha :) | 23:50 |
blackburn | sonney2k: we are doing nice progress last 1.5 months :) | 23:51 |
blackburn | I'm running out of power with improving my dim reduction 'toolkit' though | 23:51 |
@sonney2k | one of the most important things is to keep git master compilable/ tests running | 23:51 |
blackburn | I would be very disappointed being rejected with paper | 23:51 |
@sonney2k | and the debian package | 23:51 |
@sonney2k | blackburn, dont' expect that the paper will be accepted immediately | 23:52 |
blackburn | sonney2k: month? | 23:52 |
@sonney2k | there will be at least one revision | 23:52 |
@sonney2k | months + that | 23:52 |
blackburn | ah | 23:52 |
blackburn | I see | 23:53 |
blackburn | well at least after programming I can focus on paper | 23:53 |
blackburn | I feel much happier with converters concept | 23:53 |
blackburn | without any changes you can now embed features that doesn't fit in memory | 23:54 |
blackburn | sonney2k: are kernels support this? | 23:54 |
shogun-buildbot | build #288 of octave_static is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/octave_static/builds/288 | 23:54 |
@sonney2k | strike :) | 23:54 |
blackburn | nice | 23:54 |
@sonney2k | I don't understand what you are saying... | 23:55 |
blackburn | sonney2k: can I init kernel with FileFeatures or so? | 23:55 |
@sonney2k | simplefeatures don't have to be in memory... | 23:56 |
@sonney2k | so partially yes | 23:56 |
blackburn | I mean that most of methods are now working with any Features supporting distance/kernel | 23:56 |
@sonney2k | blackburn, btw the only way I see to get compilation below 1.5GB is to remove a few %template's | 23:56 |
@sonney2k | so I guess I should make them configurable | 23:57 |
@sonney2k | and drop as much as possible... | 23:57 |
@sonney2k | I see. but don't you need access to the full kernel/distance matrix? | 23:58 |
@sonney2k | or do you only access element pairs? | 23:58 |
blackburn | unfortunately I need | 23:58 |
@sonney2k | well then you shouldn't care if features fit in memory | 23:58 |
@sonney2k | these will not be the bottleneck :) | 23:59 |
blackburn | before it wasn't any possible | 23:59 |
blackburn | cause it was working only with SimpleFeatures | 23:59 |
@sonney2k | the buildbot is now working on the modular interfaces again | 23:59 |
@sonney2k | finally... | 23:59 |
@sonney2k | blackburn, I see | 23:59 |
--- Log closed Fri Oct 14 00:00:09 2011 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!