--- Log opened Sat Aug 13 00:00:18 2011 | ||
serialhex | hey blackburn whats up? | 00:24 |
---|---|---|
serialhex | aaahhhh!!! im put :P | 00:24 |
blackburn | serialhex: hi | 00:25 |
blackburn | trying to remove java serialization stuff but shogun serialization should stay %) | 00:25 |
blackburn | how are you? | 00:25 |
-!- blackburn [~blackburn@188.122.238.50] has quit [Quit: Leaving.] | 00:32 | |
-!- f-x [~user@117.192.192.210] has quit [Ping timeout: 260 seconds] | 04:25 | |
-!- in3xes [~in3xes@180.149.49.227] has joined #shogun | 05:51 | |
-!- in3xes_ [~in3xes@180.149.49.227] has joined #shogun | 06:11 | |
-!- in3xes [~in3xes@180.149.49.227] has quit [Ping timeout: 240 seconds] | 06:14 | |
-!- in3xes__ [~in3xes@180.149.49.227] has joined #shogun | 06:16 | |
-!- in3xes_ [~in3xes@180.149.49.227] has quit [Ping timeout: 240 seconds] | 06:20 | |
-!- in3xes__ is now known as in3xes | 06:21 | |
-!- in3xes [~in3xes@180.149.49.227] has quit [Client Quit] | 06:21 | |
CIA-87 | shogun: Soeren Sonnenburg master * rc1d50d1 / src/shogun/base/SGObject.h : fix compile error when compiling without reference counting - http://bit.ly/rb88nP | 07:07 |
CIA-87 | shogun: Soeren Sonnenburg master * r52fbaf1 / (3 files in 3 dirs): kill remaining warnings for python_modular - http://bit.ly/q2G7wi | 07:28 |
-!- alesis-novik [~alesis@188.74.87.206] has joined #shogun | 07:30 | |
-!- mrsrikanth [~mrsrikant@59.92.57.168] has joined #shogun | 08:26 | |
-!- blackburn [~blackburn@188.122.238.50] has joined #shogun | 09:35 | |
blackburn | sonney2k: the new problem with java serialization is that some java stuff is still here | 12:15 |
blackburn | ...sr..org.shogun.RealFeatures.OL./......xr..org.shogun.DotFeatures;A7xJM.+...xr..org.shogun.Features..^E.._....xr..org.shogun.SGObject.77...i....xpz....<<_SHOGUN_SERIALIZABLE_ASCII_FILE | 12:16 |
blackburn | and so on | 12:16 |
blackburn | that shit in the beginning should go, right? | 12:16 |
@sonney2k | blackburn, you mean in the output of the serialzed file? | 12:53 |
@sonney2k | blackburn, that is ok | 12:54 |
@sonney2k | blackburn, it is the same when we python pickle | 12:54 |
@sonney2k | blackburn, there is some $LANG specific header in tehre | 12:54 |
blackburn | but how to avoid this? | 13:01 |
blackburn | I mean we can't read java output in python | 13:01 |
blackburn | or& | 13:01 |
blackburn | or? | 13:01 |
@sonney2k | blackburn, why avoid this? | 13:06 |
@sonney2k | this is something java specific | 13:06 |
@sonney2k | I mean we use java serialization for java object | 13:06 |
@sonney2k | and python serialziation for python objects | 13:07 |
blackburn | yeah but if it is not there we could exchange objects | 13:07 |
blackburn | between java,python and everything else | 13:07 |
@sonney2k | blackburn, if we want to exchange objects we just use shogun's serialzation code | 13:07 |
@sonney2k | save_serializable() | 13:07 |
blackburn | hmm yes | 13:07 |
@sonney2k | load_serializable() | 13:07 |
blackburn | sure | 13:07 |
blackburn | okay, I'll try to get load to work now | 13:08 |
blackburn | it fails now for some strange reaso | 13:08 |
@sonney2k | the $LANG native stuff is only a convenience for people working in this $LANG | 13:08 |
blackburn | sonney2k: could you give me some advice about how to get distances between promoters? | 13:08 |
@sonney2k | I hope I will have time tonight to fix the sparse bug | 13:08 |
blackburn | or where to get comprehension of all this stuff | 13:09 |
@sonney2k | blackburn, yes look at my diss there is a tss recognition chapter in there | 13:09 |
blackburn | okay | 13:09 |
@sonney2k | and then use e.g. 2 spectrum kernels + 1 weighteddegreepositionstringkernel | 13:09 |
blackburn | how to get distance from kernel? | 13:09 |
@sonney2k | or when you want to use dotfeatures (explicit feature representatiosn) then wdfeatures / weightedspec features | 13:10 |
@sonney2k | you could use kerneldistance | 13:10 |
@sonney2k | k | 13:10 |
@sonney2k | gtg | 13:10 |
blackburn | thanks | 13:10 |
CIA-87 | shogun: Sergey Lisitsyn master * r9350b71 / src/interfaces/modular/SGBase.i : Got java externalization to work - http://bit.ly/qkjZah | 13:38 |
-!- blackburn [~blackburn@188.122.238.50] has quit [Quit: Leaving.] | 13:39 | |
-!- mrsrikanth [~mrsrikant@59.92.57.168] has quit [Quit: Leaving] | 14:21 | |
-!- mrsrikanth [~mrsrikant@59.92.57.168] has joined #shogun | 14:24 | |
-!- mrsrikanth [~mrsrikant@59.92.57.168] has quit [Quit: Leaving] | 14:34 | |
-!- blackburn [~blackburn@188.122.238.50] has joined #shogun | 14:44 | |
-!- in3xes [~in3xes@180.149.49.227] has joined #shogun | 15:44 | |
@sonney2k | blackburn, hmmhh the buildbot didn't catch your commit... | 16:19 |
blackburn | what you mean saing didn't catch? | 16:20 |
@sonney2k | blackburn, it didn't trigger any build | 16:20 |
@sonney2k | ahh exceptions.OSError: [Errno 12] Cannot allocate memory | 16:21 |
@sonney2k | that's why | 16:21 |
blackburn | sonney2k: what do you think if I rewrite PCA and KPCA and call it SimplePCA and SimpleKernelPCA | 16:22 |
blackburn | without any init and fancy options | 16:22 |
@sonney2k | why rewrite? | 16:22 |
blackburn | it takes more if I try to get into this code | 16:23 |
-!- shogun-buildbot [~shogun-bu@87.118.92.43] has quit [Quit: buildmaster reconfigured: bot disconnecting] | 16:23 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 16:23 | |
blackburn | anyway, the main point is not to use init, what do you think about it? | 16:23 |
blackburn | i.e. project vectors in apply_to_feature_matrix | 16:24 |
blackburn | like LLE, mds and other dimreduction preprocs do | 16:24 |
@sonney2k | blackburn, but why? | 16:24 |
@sonney2k | one can use pca/kpca really for preprocessing | 16:25 |
blackburn | how can one use pca in the init/apply way? | 16:25 |
@sonney2k | and e.g. the code for PCA has only a few lines... | 16:25 |
blackburn | btw PCA is broken now | 16:26 |
@sonney2k | blackburn, well you can determine the subspace and then project all vectors to come into it | 16:26 |
@sonney2k | what happened? | 16:26 |
blackburn | don't know but it acts strange | 16:26 |
@sonney2k | ? | 16:27 |
blackburn | it can't preprocess even flat grid | 16:27 |
@sonney2k | that is still to unspecific... | 16:28 |
@sonney2k | what happens? | 16:28 |
blackburn | results aren't right for sure | 16:30 |
@sonney2k | it was correct at some point ... | 16:31 |
@sonney2k | blackburn, I am working on the sparse features issue now | 16:31 |
@sonney2k | l8r | 16:31 |
blackburn | sonney2k: http://dl.dropbox.com/u/10139213/share/pca.png | 16:31 |
blackburn | and it becomes random sometimes | 16:32 |
@sonney2k | this looks like uninited memory | 16:32 |
@sonney2k | valgrind | 16:32 |
blackburn | I guess something is wrong with eigenvalues | 16:33 |
@sonney2k | it looks more like some init function or so is not called and you are getting random data | 16:34 |
@sonney2k | blackburn, python features_string_sliding_window_modular.py | 17:18 |
@sonney2k | it needs the dynamic array ... | 17:18 |
blackburn | uh | 17:18 |
blackburn | sorry didn't noticed that | 17:19 |
@sonney2k | blackburn, I've partially fixed the sparse stuff - but really we need some new way of dealing with things | 17:19 |
@sonney2k | blackburn, what I would love to have is the following | 17:20 |
@sonney2k | when one calls get_feature_vector() | 17:20 |
@sonney2k | one gets some SGVector | 17:20 |
@sonney2k | and then calls vec.free_vector() | 17:20 |
@sonney2k | that's it | 17:20 |
@sonney2k | no longer all the special stuff | 17:20 |
blackburn | ?? | 17:21 |
@sonney2k | we need the same for sparse vectors and strings | 17:21 |
@sonney2k | blackburn, did you use get_feature_vector()? | 17:21 |
blackburn | I work with feature matrices | 17:21 |
blackburn | what do you mean special stuff? | 17:22 |
-!- in3xes_ [~in3xes@180.149.49.227] has joined #shogun | 17:23 | |
@sonney2k | well you always have to call free_feature_vector() | 17:24 |
@sonney2k | same for sparse/strings | 17:25 |
-!- in3xes [~in3xes@180.149.49.227] has quit [Ping timeout: 240 seconds] | 17:26 | |
@sonney2k | blackburn, clear what I mean? | 17:32 |
@sonney2k | blackburn, btw python modelselection_grid_search_linear_modular.py | 17:32 |
@sonney2k | did you remove that too? | 17:32 |
blackburn | yeah | 17:32 |
blackburn | remove what? | 17:32 |
blackburn | ehm | 17:33 |
blackburn | nope | 17:33 |
@sonney2k | blackburn, ok then I should bug heiko about it | 17:33 |
blackburn | DynArray is the only thing to restore? | 17:33 |
CIA-87 | shogun: Soeren Sonnenburg master * raecb8bf / (4 files in 3 dirs): fix crashers in sparse features and svmocas - http://bit.ly/qmpxeA | 17:35 |
@sonney2k | DynIntArray so far only yes | 17:35 |
-!- in3xes_ is now known as in3xes | 17:37 | |
@sonney2k | shogun-buildbot, status all | 17:39 |
shogun-buildbot | cmdline_static: idle, last build 10h24m09s ago: build successful | 17:39 |
shogun-buildbot | csharp_modular: idle, last build 10h14m30s ago: build successful | 17:39 |
shogun-buildbot | java_modular: idle, last build 10h16m05s ago: build successful | 17:39 |
shogun-buildbot | libshogun: building(compiling) | 17:39 |
shogun-buildbot | lua_modular: idle, last build 10h14m26s ago: build successful | 17:39 |
shogun-buildbot | nightly_all: idle, last build 20h38m13s ago: failed configure | 17:39 |
shogun-buildbot | nightly_none: idle, last build 20h36m56s ago: failed compile | 17:39 |
shogun-buildbot | octave_modular: idle, last build 10h06m16s ago: build successful | 17:39 |
shogun-buildbot | octave_static: idle, last build 10h24m00s ago: build successful | 17:39 |
shogun-buildbot | python_modular: idle, last build 10h07m04s ago: build successful | 17:39 |
shogun-buildbot | python_static: idle, last build 10h24m07s ago: build successful | 17:39 |
shogun-buildbot | r_modular: idle, last build 10h07m05s ago: build successful | 17:39 |
shogun-buildbot | r_static: idle, last build 10h24m10s ago: build successful | 17:39 |
shogun-buildbot | ruby_modular: idle, last build 10h15m59s ago: build successful | 17:39 |
@sonney2k | hmmhh now it is building... | 17:39 |
@sonney2k | blackburn, what did you do to him that he ignored you? | 17:39 |
blackburn | sonney2k: I don't know, may be he doesn't like russians | 17:40 |
blackburn | :D | 17:40 |
@sonney2k | must be memory issues | 17:40 |
-!- in3xes [~in3xes@180.149.49.227] has quit [Ping timeout: 240 seconds] | 18:06 | |
-!- in3xes [~in3xes@180.149.49.227] has joined #shogun | 18:09 | |
blackburn | sonney2k: kernel pca init is a little bit tricky, have you finished it? | 19:36 |
@sonney2k | blackburn, kernel pca init works | 20:10 |
@sonney2k | the problem is that apply does not | 20:10 |
-!- in3xes [~in3xes@180.149.49.227] has quit [Ping timeout: 240 seconds] | 20:11 | |
blackburn | ясно | 20:17 |
CIA-87 | shogun: Soeren Sonnenburg master * rad1314a / (2 files in 2 dirs): fix remaining warnings in libshogun - http://bit.ly/pYmTBW | 20:21 |
@sonney2k | blackburn, look just 3 warnings http://shogun-toolbox.org:5347/waterfall | 20:22 |
@sonney2k | blackburn, regarding kernelpca | 20:22 |
blackburn | 3 warnings left?! | 20:22 |
@sonney2k | I ported what vojtech uses | 20:22 |
@sonney2k | blackburn, in libshogun | 20:22 |
blackburn | sonney2k: I'm looking at it now | 20:22 |
blackburn | can't understand how to project/apply data | 20:22 |
@sonney2k | blackburn, do you have the matlab file / mex file? | 20:23 |
blackburn | it seems it involves different routine | 20:23 |
blackburn | I have one from his statistical pr toolbox | 20:23 |
@sonney2k | I did compare the result of the kpca components - they where the same | 20:23 |
@sonney2k | blackburn, exactly that is what I am using | 20:23 |
@sonney2k | err was using | 20:23 |
blackburn | sonney2k: so kpcarec should be used as mapping to new feature space? | 20:24 |
-!- f-x [~user@117.192.212.140] has joined #shogun | 20:33 | |
blackburn | sonney2k: so you killed out all the doxygen warnings? | 20:35 |
@sonney2k | 0 warnings http://shogun-toolbox.org:5347/waterfall! | 20:38 |
@sonney2k | blackburn, no | 20:38 |
@sonney2k | doxygen ones are still there | 20:39 |
@sonney2k | f-x, I hopefullly correctly fixed the vw warning https://github.com/shogun-toolbox/shogun/commit/ad1314a98b0c84ad32e19f0485cfc8c3c182a003 | 20:39 |
f-x | sonney2k: yeah, i checked it out.. looks nice | 20:40 |
@sonney2k | blackburn, I don't remember the variable names - in fact I forgot everything about kpca (well almost ;) | 20:40 |
@sonney2k | f-x, I think it was wrong before even ... | 20:40 |
@sonney2k | the char* x[total] | 20:40 |
blackburn | sonney2k: not variable names, which .m routine makes a projection | 20:41 |
f-x | sonney2k: wrong? in that it wouldn't work? | 20:41 |
@sonney2k | f-x, that would be total char* pointers | 20:41 |
f-x | oh - the allocation of the array? | 20:41 |
@sonney2k | so not just one size_t variable | 20:41 |
@sonney2k | blackburn, I only recall that it called some mex function to apply kpca | 20:41 |
blackburn | okay | 20:42 |
f-x | sonney2k: yes, it should have been char x[total] | 20:44 |
@sonney2k | yeah or just size_t x :) | 20:44 |
f-x | exactly :) | 20:44 |
f-x | i'll put in the actual learning algorithm code soon, hopefully that won't produce any warnings | 20:45 |
f-x | sonney2k: do we have the feature of loading a saved learner state? | 20:45 |
f-x | and then continuing from that? | 20:45 |
@sonney2k | f-x, not in general no | 20:45 |
@sonney2k | svmlight can do sth like this when alphas are set | 20:45 |
@sonney2k | for online learners / liblinear that would be relatively easy to add but it is not there yet | 20:46 |
f-x | sonney2k: hmm i was wondering if there's a general method to do this, or if i should do it in the VW-specific way for now | 20:46 |
f-x | ok, so now i'll do it how they do it for VW | 20:46 |
@sonney2k | yes | 20:47 |
f-x | but not compatible with other online learners | 20:47 |
@sonney2k | blackburn, that are the remaining warnings for python_modular - ~1200 and pretty much all documentation http://shogun-toolbox.org:5347/builders/python_modular/builds/6/steps/compile/logs/warnings%20(1240)/text - search for preprocessor and you will find enough to fix :) | 20:49 |
@sonney2k | f-x, and you search for vw in http://shogun-toolbox.org:5347/builders/python_modular/builds/6/steps/compile/logs/warnings%20(1240)/text for warnings | 20:50 |
@sonney2k | to fix | 20:50 |
@sonney2k | undocumented stuff | 20:50 |
f-x | yeah, sure.. that's the next task | 20:53 |
blackburn | sonney2k: serialization of libsvm in java fails | 21:00 |
blackburn | with memory corruption | 21:00 |
blackburn | I guess not my fault | 21:00 |
-!- f-x [~user@117.192.212.140] has quit [Remote host closed the connection] | 21:00 | |
@sonney2k | blackburn, doesn't it fail for everything? | 21:00 |
blackburn | no, serialization in java works for labels, features and kernels for sure | 21:01 |
CIA-87 | shogun: Sergey Lisitsyn master * r86d2c7f / src/interfaces/modular/Library.i : Restored Dynarray modular renames - http://bit.ly/pNTBJJ | 21:03 |
CIA-87 | shogun: Sergey Lisitsyn master * r0f67ca8 / src/interfaces/modular/SGBase.i : Codestyle fix for java externalization routines - http://bit.ly/nGysFr | 21:03 |
CIA-87 | shogun: Sergey Lisitsyn master * re57e521 / examples/undocumented/java_modular/serialization_svmlight_modular.java : Improved java svmlight externalization example - http://bit.ly/q7np0B | 21:03 |
blackburn | sonney2k: you could check by yourself with examples/undocumented/java_modular/serialization_svmlight_modular.java | 21:03 |
blackburn | shogun-buildbot: pull the trigger | 21:03 |
@sonney2k | blackburn, that fails too here | 21:04 |
@sonney2k | in pythoon modular | 21:05 |
@sonney2k | hmmhh | 21:05 |
blackburn | boom | 21:05 |
@sonney2k | rebooted | 21:05 |
@sonney2k | well out of memory | 21:05 |
@sonney2k | we need fatter buildbots | 21:06 |
@sonney2k | too bad that I couldn't plug in 16G ... | 21:06 |
blackburn | sonney2k: I guess we should ignore moreee | 21:07 |
shogun-buildbot | build #1 of nightly_none is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org:5347/builders/nightly_none/builds/1 | 21:07 |
@sonney2k | hurray | 21:07 |
@sonney2k | the full one of course failed because I have no cplex here | 21:07 |
blackburn | me too | 21:09 |
blackburn | sonney2k: now we have 616 classes | 21:11 |
blackburn | in interfaces | 21:11 |
-!- in3xes [~in3xes@180.149.49.227] has joined #shogun | 21:12 | |
blackburn | sonney2k: what is Plif? | 21:12 |
@sonney2k | blackburn, piecewise linear function | 21:12 |
@sonney2k | needed in structured output stuff | 21:12 |
blackburn | oooh | 21:12 |
blackburn | bad thing I can't see any class to remove | 21:13 |
-!- f-x [~user@117.192.212.140] has joined #shogun | 21:20 | |
blackburn | sonney2k: may be reducing number of different types could reduce memory usage | 21:26 |
blackburn | seriously, who ever wants to use uint8_t | 21:27 |
@sonney2k | blackburn, certainly | 21:27 |
@sonney2k | I use it all the time for strings | 21:27 |
blackburn | but simplefeatures? | 21:27 |
@sonney2k | and think of images | 21:27 |
blackburn | so your point is we can't remove any? | 21:28 |
blackburn | sonney2k: hmm found one way | 21:30 |
blackburn | sonney2k: swig creates wrappers for "short", etc | 21:30 |
blackburn | bool, clock_t | 21:31 |
blackburn | I guess we can avoid this | 21:32 |
blackburn | don't sure yet | 21:32 |
-!- f-x [~user@117.192.212.140] has quit [Remote host closed the connection] | 21:33 | |
@sonney2k | clock_t? | 21:33 |
blackburn | yeah, it creates wrappers for clock_t too | 21:34 |
blackburn | without renaming though | 21:34 |
blackburn | hmm no | 21:36 |
blackburn | damn it | 21:36 |
shogun-buildbot | build #8 of octave_modular is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org:5347/builders/octave_modular/builds/8 blamelist: blackburn91@gmail.com | 21:45 |
CIA-87 | shogun: Sergey Lisitsyn master * rdbf2ab3 / (5 files): Got rid of Preprocessor doxygen warnings - http://bit.ly/pBq6Zj | 21:45 |
blackburn | hehe | 21:46 |
-!- alesis-novik [~alesis@188.74.87.206] has quit [Quit: Leaving] | 21:47 | |
blackburn | shogun-buildbot: status | 21:53 |
shogun-buildbot | cmdline_static: building(compiling) [ETA 15m35s] | 21:53 |
shogun-buildbot | csharp_modular: building(compiling) [ETA 5m00s] | 21:53 |
shogun-buildbot | java_modular: building(compiling) [ETA 5m22s] | 21:53 |
shogun-buildbot | libshogun: idle, last build 1m37s ago: build successful | 21:53 |
shogun-buildbot | lua_modular: building(compiling) [ETA 5m02s] | 21:53 |
shogun-buildbot | nightly_all: idle, last build 53m14s ago: failed configure | 21:53 |
shogun-buildbot | nightly_none: idle, last build 46m10s ago: build successful | 21:53 |
shogun-buildbot | octave_modular: building(compiling) [ETA 15m05s] | 21:53 |
shogun-buildbot | octave_static: building(compiling) [ETA 8m41s] | 21:53 |
shogun-buildbot | python_modular: building(compiling) [ETA 2 seconds], (compiling) [ETA 11m26s] | 21:53 |
shogun-buildbot | python_static: building(compiling) [ETA 8m34s] | 21:53 |
shogun-buildbot | r_modular: building(compiling) [ETA 9 seconds], (compiling) [ETA 9 seconds] | 21:53 |
shogun-buildbot | r_static: building(compiling) [ETA 15m48s] | 21:53 |
shogun-buildbot | ruby_modular: building(compiling) [ETA 4m59s] | 21:53 |
-!- f-x [~user@117.192.212.140] has joined #shogun | 21:57 | |
shogun-buildbot | build #10 of python_modular is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org:5347/builders/python_modular/builds/10 | 22:27 |
blackburn | 1220 | 22:27 |
shogun-buildbot | build #11 of octave_modular is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org:5347/builders/octave_modular/builds/11 | 22:29 |
CIA-87 | shogun: Sergey Lisitsyn master * rb1d1a6b / (src/shogun/base/Parameter.h src/shogun/base/Version.h): Fixed more warnings (Parameter,Version) - http://bit.ly/pGn1YX | 22:59 |
-!- alesis-novik [~alesis@cpat001.wlan.net.ed.ac.uk] has joined #shogun | 23:15 | |
blackburn | 982 | 23:26 |
-!- blackburn [~blackburn@188.122.238.50] has quit [Quit: Leaving.] | 23:33 | |
CIA-87 | shogun: Shashwat Lal Das master * r3b4d080 / (4 files): Added dump_regressor to VwRegressor. - http://bit.ly/p8REQb | 23:42 |
CIA-87 | shogun: Shashwat Lal Das master * r94fd8ae / (2 files): Made VW version part of env. - http://bit.ly/oY5IM3 | 23:42 |
CIA-87 | shogun: Shashwat Lal Das master * r5d60b6a / src/shogun/classifier/vw/VwRegressor.cpp : Minor type change in VwRegressor. - http://bit.ly/nEr9R7 | 23:42 |
CIA-87 | shogun: Soeren Sonnenburg master * r60e3bad / (6 files in 2 dirs): | 23:42 |
CIA-87 | shogun: Merge pull request #293 from frx/streaming_vw | 23:42 |
CIA-87 | shogun: dump_regressor added to VwRegressor - http://bit.ly/nLN9uo | 23:42 |
-!- in3xes [~in3xes@180.149.49.227] has quit [Ping timeout: 240 seconds] | 23:45 | |
--- Log closed Sun Aug 14 00:00:28 2011 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!