--- Log opened Sat May 18 00:00:06 2013 | ||
-!- nube [~rho@110.44.115.229] has quit [Quit: Leaving.] | 00:21 | |
-!- dvalcarce [5b757556@gateway/web/freenode/ip.91.117.117.86] has quit [Ping timeout: 250 seconds] | 00:46 | |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 00:56 | |
-!- van51 [~van51@athedsl-320452.home.otenet.gr] has quit [Ping timeout: 260 seconds] | 01:30 | |
-!- foulwall [~foulwall@2001:da8:215:503:7592:c803:cf8a:68f5] has joined #shogun | 03:27 | |
shogun-buildbot | build #398 of nightly_default is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_default/builds/398 | 03:32 |
---|---|---|
-!- foulwall [~foulwall@2001:da8:215:503:7592:c803:cf8a:68f5] has quit [Remote host closed the connection] | 04:18 | |
-!- foulwall [~user@2001:da8:215:503:7592:c803:cf8a:68f5] has joined #shogun | 04:19 | |
-!- votjakovr [~votjakovr@host-46-241-3-209.bbcustomer.zsttk.net] has joined #shogun | 05:46 | |
-!- foulwall [~user@2001:da8:215:503:7592:c803:cf8a:68f5] has quit [Ping timeout: 264 seconds] | 05:52 | |
-!- nube [~rho@49.244.56.213] has joined #shogun | 06:01 | |
-!- foulwall [~user@2001:da8:215:503:7592:c803:cf8a:68f5] has joined #shogun | 06:06 | |
-!- foulwall [~user@2001:da8:215:503:7592:c803:cf8a:68f5] has quit [Ping timeout: 264 seconds] | 08:17 | |
-!- foulwall [~user@2001:da8:215:6901:f0f2:469f:ab0:2494] has joined #shogun | 08:19 | |
-!- foulwall [~user@2001:da8:215:6901:f0f2:469f:ab0:2494] has quit [Ping timeout: 245 seconds] | 08:26 | |
-!- foulwall [~user@2001:da8:215:503:81a:d016:1e9d:4fda] has joined #shogun | 08:40 | |
-!- iglesiasg [~fjig@share-01.csc.kth.se] has joined #shogun | 09:10 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 09:10 | |
-!- hushell [~hushell@c-67-189-100-116.hsd1.or.comcast.net] has joined #shogun | 09:36 | |
-!- vgorbati [4d2f813b@gateway/web/freenode/ip.77.47.129.59] has joined #shogun | 09:48 | |
-!- hoijui [~hoijui@dslb-088-074-126-103.pools.arcor-ip.net] has joined #shogun | 09:54 | |
foulwall | morning everyone! | 10:01 |
@iglesiasg | good morning | 10:01 |
foulwall | Hi iglesiasg! | 10:02 |
-!- vgorbati [4d2f813b@gateway/web/freenode/ip.77.47.129.59] has quit [Ping timeout: 250 seconds] | 10:14 | |
@iglesiasg | SystemError: Out of memory error, tried to allocate -6464406072 bytes using malloc. | 10:19 |
@iglesiasg | discovering new errors | 10:19 |
@iglesiasg | O_O | 10:19 |
@lisitsyn | iglesiasg: sounds like overflow! | 10:45 |
@iglesiasg | lisitsyn: I have a bug for sure, something uninitialized I'd bet | 10:52 |
@iglesiasg | sssssssssss' | 11:36 |
@iglesiasg | oops sorry | 11:36 |
-!- van511 [~van51@athedsl-320452.home.otenet.gr] has joined #shogun | 11:48 | |
van511 | hello | 12:04 |
-!- van511 is now known as van51 | 12:04 | |
van51 | sonney2k: I made a PR, about that insertion issue and now I am waiting for travis :D | 12:13 |
-!- foulwall [~user@2001:da8:215:503:81a:d016:1e9d:4fda] has quit [Ping timeout: 260 seconds] | 12:14 | |
van51 | sonney2k: so it turns out that, although it was doing an invalid memory read every time the insert was called on an empty array, it did add that element and everything was working ok after that | 12:30 |
van51 | the problem was in a loop in CombinedKernel where it went over all the kernels of the array and all the features | 12:31 |
van51 | and it wasn't supposed to increment the feature index if the kernel was a CustomKernel | 12:31 |
van51 | that's why it crashed in python and not in octave | 12:31 |
van51 | because the example in python used a CustomKernel | 12:31 |
-!- hoijui [~hoijui@dslb-088-074-126-103.pools.arcor-ip.net] has quit [Quit: Leaving] | 12:56 | |
-!- foulwall [~foulwall@li379-21.members.linode.com] has joined #shogun | 12:56 | |
@sonney2k | van51, excellent analysis :D | 14:18 |
@sonney2k | foulwall, have a look I slightly tried to improve the gp example | 14:18 |
@sonney2k | foulwall, ohh and I created the repository under shogun-toolbox | 14:19 |
@sonney2k | iglesiasg, heh! | 14:19 |
-!- foulwall [~foulwall@li379-21.members.linode.com] has quit [Ping timeout: 252 seconds] | 14:20 | |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 14:21 | |
shogun-notifier- | shogun: van51 :develop * af89c11 / / (6 files): https://github.com/shogun-toolbox/shogun/commit/af89c11e5061d40732a7ff3600f6ae8b64dbcc2d | 14:21 |
shogun-notifier- | shogun: Fixed insert in CombinedKernel/CombinedFeatures -- Added some unit tests | 14:21 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * 8b54eaf / / (6 files): https://github.com/shogun-toolbox/shogun/commit/8b54eaff29b7197bcb5f862aabb11b26d0595c12 | 14:21 |
shogun-notifier- | shogun: Merge pull request #1108 from van51/feature/fix_insert_bug | 14:21 |
shogun-notifier- | shogun: | 14:21 |
shogun-notifier- | shogun: Fixed insertion in CombinedKernel/CombinedFeatures | 14:21 |
shogun-buildbot | build #990 of deb2 - static_interfaces is complete: Failure [failed test libshogun] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/990 blamelist: Soeren Sonnenburg <sonne@debian.org> | 14:31 |
shogun-buildbot | build #999 of bsd1 - libshogun is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/999 | 14:34 |
-!- foulwall_ [~foulwall@2001:da8:215:503:845d:8a7c:4625:8019] has joined #shogun | 14:37 | |
foulwall_ | sonney2k: new Gp-demo not working, trying to find the bug. | 14:39 |
-!- travis-ci [~travis-ci@ec2-23-22-169-12.compute-1.amazonaws.com] has joined #shogun | 14:41 | |
travis-ci | [travis-ci] it's Soeren Sonnenburg's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/7275889 | 14:41 |
-!- travis-ci [~travis-ci@ec2-23-22-169-12.compute-1.amazonaws.com] has left #shogun [] | 14:41 | |
shogun-buildbot | build #836 of cyg1 - libshogun is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/836 blamelist: van51 <vangelis_51@hotmail.com> | 14:44 |
shogun-buildbot | build #991 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/991 | 14:47 |
shogun-buildbot | build #1129 of deb3 - modular_interfaces is complete: Failure [failed test libshogun] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1129 blamelist: Soeren Sonnenburg <sonne@debian.org> | 15:08 |
-!- iglesiasg [~fjig@share-01.csc.kth.se] has quit [Quit: leaving] | 15:08 | |
shogun-buildbot | build #1130 of deb3 - modular_interfaces is complete: Failure [failed test python_modular] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1130 blamelist: van51 <vangelis_51@hotmail.com> | 15:27 |
-!- foulwall_ [~foulwall@2001:da8:215:503:845d:8a7c:4625:8019] has quit [Remote host closed the connection] | 16:12 | |
-!- foulwall [~foulwall@2001:da8:215:c252:88d0:3a67:5118:5696] has joined #shogun | 17:09 | |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 17:21 | |
@lisitsyn | sonney2k: alright really no chance to get another slot | 17:35 |
@lisitsyn | they would want that excellent outstanding proposal we need a slot for | 17:36 |
-!- foulwall [~foulwall@2001:da8:215:c252:88d0:3a67:5118:5696] has quit [Remote host closed the connection] | 17:46 | |
-!- vgorbati [d4029f22@gateway/web/freenode/ip.212.2.159.34] has joined #shogun | 18:16 | |
-!- vgorbati [d4029f22@gateway/web/freenode/ip.212.2.159.34] has quit [Ping timeout: 250 seconds] | 19:18 | |
-!- nube [~rho@49.244.56.213] has quit [Ping timeout: 248 seconds] | 19:22 | |
-!- van51 [~van51@athedsl-320452.home.otenet.gr] has quit [Ping timeout: 256 seconds] | 19:26 | |
-!- van51 [~van51@athedsl-320452.home.otenet.gr] has joined #shogun | 19:33 | |
-!- nube [~rho@49.244.110.205] has joined #shogun | 19:35 | |
-!- vgorbati [4e6fbbaa@gateway/web/freenode/ip.78.111.187.170] has joined #shogun | 19:51 | |
@sonney2k | lisitsyn, of course not | 20:01 |
@sonney2k | if we had 2 slots | 20:01 |
@sonney2k | then maybe we woudl be considered | 20:01 |
@lisitsyn | sonney2k: yeah we must be happy with that crazy amount | 20:02 |
@lisitsyn | we have now | 20:02 |
@sonney2k | exactly | 20:03 |
@sonney2k | being greedy is not a good thing | 20:03 |
@sonney2k | better fewer students and fun than maxxed out to the limit | 20:06 |
@sonney2k | van51, we are getting closer :-) | 20:06 |
@sonney2k | van51, the buildbot at least for static interfaces is happy again | 20:06 |
@lisitsyn | sonney2k: I am now getting some expertise in opencl | 20:07 |
@lisitsyn | have to | 20:07 |
@sonney2k | lisitsyn, is there any ML algorithm except NN's benefiting from that? | 20:09 |
@lisitsyn | sonney2k: I didn't try to implement anything | 20:10 |
@lisitsyn | sonney2k: well DR is implementable with gpus | 20:10 |
@lisitsyn | sonney2k: svms - you may know better | 20:10 |
@sonney2k | I tried once - not with opencl but cuda but it wasn't worth it | 20:10 |
@lisitsyn | well but you are not an cuda expert | 20:10 |
@lisitsyn | that matters a lot | 20:10 |
@lisitsyn | so I don't know | 20:11 |
@sonney2k | for svms maybe some block wise thing would work | 20:11 |
@sonney2k | lisitsyn, surely not but you cannot a single bit improve if you have random memory access patterns | 20:11 |
@sonney2k | and I had that | 20:11 |
@lisitsyn | sonney2k: yes, I expect some mini batch here instead | 20:11 |
@sonney2k | and worse | 20:12 |
@sonney2k | you need to access memory linearly | 20:13 |
@sonney2k | that works fine for e.g. gaussian kernel and others | 20:13 |
@lisitsyn | sonney2k: I mean batched SGD would work really fine here | 20:13 |
@sonney2k | sure | 20:13 |
@sonney2k | is it worth the effort though? | 20:14 |
@lisitsyn | sonney2k: I don't know - some-x speedup | 20:14 |
van51 | sonney2k: what is it unhappy for? :) | 20:15 |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 20:15 | |
shogun-notifier- | shogun: iglesias :develop * c63fee3 / src/shogun/structure/ (2 files): https://github.com/shogun-toolbox/shogun/commit/c63fee3820fac7bf32bedf2ade966f22c1a91866 | 20:15 |
shogun-notifier- | shogun: Fix SO-SVM constructors missing call to init. | 20:15 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * 5bd41d9 / src/shogun/structure/ (2 files): https://github.com/shogun-toolbox/shogun/commit/5bd41d941fe4b3d7475fad2bef16d95f24236c06 | 20:15 |
shogun-notifier- | shogun: Merge pull request #1109 from iglesias/feature/fix_init | 20:15 |
shogun-notifier- | shogun: | 20:15 |
shogun-notifier- | shogun: Fix SO-SVM constructors missing call to init. | 20:15 |
@sonney2k | van51, some tests still fail | 20:16 |
@sonney2k | van51, I am investigating | 20:16 |
@sonney2k | foulwall - gp example works fine here | 20:16 |
vgorbati | sonney2k: hello, I have just sent a PR for the issue 'remove CFeatures::get_size' | 20:16 |
van51 | ah I see some it failed inside python modular | 20:16 |
vgorbati | sonney2k: no surpises there, I've just removed all get_size methods and that's all:) | 20:17 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * 8390dae / src/shogun/regression/GaussianProcessRegression.cpp: https://github.com/shogun-toolbox/shogun/commit/8390daeacb2aba2b0c7462e1e01c51c81565b239 | 20:20 |
shogun-notifier- | shogun: add whitespace | 20:20 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * 7491998 / examples/undocumented/libshogun/regression_gaussian_process_sum.cpp: https://github.com/shogun-toolbox/shogun/commit/7491998e54c890217b39cfc72ca22e65c620478c | 20:20 |
shogun-notifier- | shogun: temporarily disable combined kernels in gp example | 20:20 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * 5948cc0 / src/configure: https://github.com/shogun-toolbox/shogun/commit/5948cc0a05c149b0ac4c8208dba4c49d1785ef6a | 20:20 |
shogun-notifier- | shogun: mention prng for mersenne | 20:20 |
shogun-notifier- | shogun: vladislav.horbatiuk@gmail.com :develop * a633ff6 / src/shogun/features/ (45 files): https://github.com/shogun-toolbox/shogun/commit/a633ff6f4d9fd9a7672144bee73bc1c5e52938bb | 20:21 |
shogun-notifier- | shogun: Removed CFeatures::get_size() method (according to https://github.com/shogun-toolbox/shogun/issues/1102). | 20:21 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * 436faf0 / src/shogun/features/ (45 files): https://github.com/shogun-toolbox/shogun/commit/436faf037d52b08ce62c5dd86f6e926bb9e67e0b | 20:21 |
shogun-notifier- | shogun: Merge pull request #1112 from vladislav-horbatiuk/vgorbati-develop | 20:21 |
shogun-notifier- | shogun: | 20:21 |
shogun-notifier- | shogun: Removed CFeatures::get_size() method (Closes: #1112) | 20:21 |
@lisitsyn | sonney2k: what about one more infrastructure thing for vgorbati? | 20:24 |
shogun-buildbot | build #992 of deb2 - static_interfaces is complete: Failure [failed test libshogun] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/992 blamelist: Soeren Sonnenburg <sonne@debian.org> | 20:24 |
@sonney2k | lisitsyn, what do I have to write to get an issue closed in a git commit comment? | 20:25 |
@lisitsyn | sonney2k: closes #666 | 20:25 |
@sonney2k | so Closes : #xxx won't work? | 20:30 |
@sonney2k | lisitsyn, is that documented somehwere? | 20:31 |
@lisitsyn | sonney2k: yes | 20:31 |
@lisitsyn | sonney2k: https://github.com/blog/1386-closing-issues-via-commit-messages | 20:31 |
@lisitsyn | sonney2k: close closes closed fixes fixed | 20:31 |
van51 | sonney2k: can I assist in the investigation? :P | 20:32 |
-!- travis-ci [~travis-ci@ec2-107-22-88-191.compute-1.amazonaws.com] has joined #shogun | 20:35 | |
travis-ci | [travis-ci] it's Soeren Sonnenburg's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/7282608 | 20:35 |
-!- travis-ci [~travis-ci@ec2-107-22-88-191.compute-1.amazonaws.com] has left #shogun [] | 20:35 | |
shogun-buildbot | build #837 of cyg1 - libshogun is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/837 | 20:35 |
shogun-buildbot | build #838 of cyg1 - libshogun is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/838 blamelist: iglesias <fernando.iglesiasg@gmail.com> | 20:36 |
@sonney2k | van51, I am just running make tests locally and am checking | 20:38 |
@sonney2k | van51, once I see sth I will let you know - thanks! | 20:38 |
@sonney2k | lisitsyn, ohh we can change the default branch too | 20:39 |
@sonney2k | lisitsyn, cool faq :D | 20:39 |
shogun-buildbot | build #839 of cyg1 - libshogun is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/839 blamelist: Soeren Sonnenburg <sonne@debian.org>, vladislav.horbatiuk@gmail.com <vladislav.horbatiuk@gmail.com> | 20:40 |
vgorbati | sonney2k: is the issue https://github.com/shogun-toolbox/shogun/issues/906 still valid? | 20:42 |
-!- votjakovr [~votjakovr@host-46-241-3-209.bbcustomer.zsttk.net] has quit [Quit: Leaving] | 20:44 | |
shogun-buildbot | build #1131 of deb3 - modular_interfaces is complete: Failure [failed test libshogun] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1131 blamelist: Soeren Sonnenburg <sonne@debian.org> | 20:50 |
@sonney2k | vgorbati, yeah but that is tough | 20:51 |
-!- travis-ci [~travis-ci@ec2-50-16-137-229.compute-1.amazonaws.com] has joined #shogun | 20:51 | |
travis-ci | [travis-ci] it's Soeren Sonnenburg's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/7282706 | 20:51 |
-!- travis-ci [~travis-ci@ec2-50-16-137-229.compute-1.amazonaws.com] has left #shogun [] | 20:51 | |
@sonney2k | van51, btw what is missing now? product kernel / combined kernel / combined features are all gone now or? | 20:51 |
@sonney2k | van51, I mean wrt CList -> Dynarray | 20:52 |
van51 | sonney2k: I am currently working on ProductKernel | 20:52 |
shogun-buildbot | build #993 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/993 | 20:52 |
van51 | sonney2k: I have one method left in that class and then some small changes in other classes that reference it | 20:52 |
-!- vgorbati [4e6fbbaa@gateway/web/freenode/ip.78.111.187.170] has quit [Ping timeout: 250 seconds] | 20:53 | |
@sonney2k | van51, ok keep going then | 20:53 |
@sonney2k | man I screwed tester.py | 20:54 |
@sonney2k | it was going printing berserk | 20:54 |
@sonney2k | van51, ok we have 3 tests failing related to combined kernel: kernel_combined_custom_poly_modular.py mkl_binclass_modular.py statistics_mmd_kernel_selection_combined.py | 20:55 |
@sonney2k | van51, ohh but it is pretty clear | 20:56 |
@sonney2k | it cannot read in the old test data because lhs/feature_array etc didn't exist yet | 20:56 |
@sonney2k | van51, so I only have to regenerate integration test data and then it should be all good | 20:58 |
van51 | sonney2k: ok! let me know | 20:58 |
van51 | sonney2k: you regenerate them throught generate_testdata.py? | 20:59 |
van51 | through* | 20:59 |
@sonney2k | van51, just generator.py in python_modular dir | 21:02 |
@sonney2k | van51, that will then in shogun/data/tests create the 'references' | 21:03 |
@sonney2k | for the regression tests | 21:03 |
-!- travis-ci [~travis-ci@ec2-107-22-88-191.compute-1.amazonaws.com] has joined #shogun | 21:08 | |
travis-ci | [travis-ci] it's Soeren Sonnenburg's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/7282777 | 21:08 |
-!- travis-ci [~travis-ci@ec2-107-22-88-191.compute-1.amazonaws.com] has left #shogun [] | 21:08 | |
van51 | sonney2k: I see the files.. are these serializations of the expected values? | 21:09 |
@sonney2k | van51, the way we do examples is that they are all functions returing some values | 21:11 |
@sonney2k | (at least for python_modular) | 21:11 |
@sonney2k | so what is returned is serialized | 21:11 |
@sonney2k | in python slang pickle'd | 21:11 |
shogun-notifier- | shogun-data: Soeren Sonnenburg :master * 10d6bd9 / testsuite/tests/ (7 files): https://github.com/shogun-toolbox/shogun-data/commit/10d6bd96d3a99c22d5e6863abfb1b2f7dc69582d | 21:12 |
shogun-notifier- | shogun-data: update tests using combined kernel / features for list -> array change | 21:12 |
van51 | sonney2k: and then the output from edited code is compared against those serializations? | 21:13 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * 829c74a / data,tests/integration/python_modular/tester.py: https://github.com/shogun-toolbox/shogun/commit/829c74acfc69fb2c5f238e4d87da3d286b5b9ac8 | 21:13 |
shogun-notifier- | shogun: fix utterly verbose tester and require new data version | 21:13 |
@sonney2k | van51, yes tester.py is just doing that | 21:13 |
@sonney2k | it calls the function gets the output and loads the serialized stuff and does a 1:1 string comparison | 21:14 |
van51 | sonney2k: so you take for granted that the moment when you generate those files the code works fine | 21:14 |
van51 | sonney2k: I am trying to understand, now that you regenerated those files to have the appropriate variable name, won't they also have as correct output the output from my version? | 21:16 |
@sonney2k | van51, exactly | 21:16 |
van51 | so even if it isn't bug-free, it can't fail | 21:17 |
@sonney2k | van51, well you checked that already with the octave example | 21:17 |
@sonney2k | van51, and your (impressive!) unit tests make me very confident that what you did is fine | 21:18 |
van51 | sonney2k: I sense a little irony there? :P | 21:18 |
@sonney2k | van51, no! | 21:20 |
@sonney2k | I mean it! | 21:20 |
van51 | sonney2k: maybe in situations like these, you could just sed the generated file | 21:20 |
@sonney2k | they are realy thourough | 21:20 |
van51 | so that the number outputs remain from the original | 21:20 |
@sonney2k | van51, yeah but not easy in this case | 21:20 |
@sonney2k | van51, to make your mind at peace | 21:20 |
@sonney2k | I just checked the diff | 21:20 |
@sonney2k | output *is* the same | 21:20 |
@sonney2k | van51, btw we print a diff of the serialized vs. run outputs | 21:21 |
@sonney2k | so that is how I know | 21:21 |
@sonney2k | and it displays only the feature stuff | 21:21 |
van51 | sonney2k: ok! got it | 21:23 |
van51 | I also wanted to understand the way that testing worked | 21:24 |
@sonney2k | lisitsyn, any idea why https://travis-ci.org/shogun-toolbox/shogun/jobs/7283797 IsomapTest.neighbors_preserving suddenly fails? | 21:24 |
van51 | and also if you're at ease, then I'm at ease :) | 21:24 |
@lisitsyn | sonney2k: argh | 21:24 |
@lisitsyn | forgot to take a look | 21:24 |
@lisitsyn | sonney2k: no I can't reproduce it actually | 21:24 |
@sonney2k | lisitsyn, it fails only with clang! | 21:24 |
@lisitsyn | let's disable it | 21:24 |
@lisitsyn | hmmmmm | 21:24 |
@lisitsyn | always with clang? | 21:24 |
@sonney2k | IDK | 21:25 |
@sonney2k | guess not | 21:25 |
shogun-buildbot | build #840 of cyg1 - libshogun is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/840 blamelist: Soeren Sonnenburg <sonne@debian.org> | 21:25 |
@lisitsyn | let me try | 21:25 |
@sonney2k | van51, well what we do for testing is: | 21:25 |
@sonney2k | a) run unit tests | 21:25 |
@sonney2k | b) run all examples (jsut run no checks) | 21:25 |
@sonney2k | c) run legacy integration tests (they are hefty hand work not easy to extend but we keep them around since they are the only way to test static interfaces currently) | 21:26 |
@sonney2k | d) run the new integration tests I just described | 21:26 |
@sonney2k | only problem currently is that graphical tests are not run | 21:27 |
@sonney2k | they would fail since no X / interactive | 21:27 |
shogun-buildbot | build #1132 of deb3 - modular_interfaces is complete: Failure [failed test python_modular] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1132 blamelist: Soeren Sonnenburg <sonne@debian.org>, iglesias <fernando.iglesiasg@gmail.com>, vladislav.horbatiuk@gmail.com <vladislav.horbatiuk@gmail.com> | 21:30 |
van51 | so it's still failing :S | 21:32 |
van51 | or it is an old one? | 21:32 |
-!- travis-ci [~travis-ci@ec2-23-22-169-12.compute-1.amazonaws.com] has joined #shogun | 21:34 | |
travis-ci | [travis-ci] it's Soeren Sonnenburg's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/7283795 | 21:34 |
-!- travis-ci [~travis-ci@ec2-23-22-169-12.compute-1.amazonaws.com] has left #shogun [] | 21:34 | |
@sonney2k | van51, give it time | 21:38 |
@sonney2k | in 28 mins we will know more | 21:39 |
van51 | ok, getting back to work then :D | 21:40 |
@sonney2k | wiking, look at these warnings - http://shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1133/steps/compile%20python_modular/logs/warnings%20%2825%29 | 21:55 |
@sonney2k | wiking, many are coused by SFMT - could you please fix them? | 21:56 |
shogun-buildbot | build #1133 of deb3 - modular_interfaces is complete: Failure [failed test python_modular] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1133 blamelist: Soeren Sonnenburg <sonne@debian.org> | 21:57 |
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has quit [Ping timeout: 248 seconds] | 21:57 | |
@sonney2k | van51, everything works except for evaluation_multiclassovrevaluation_modular.py | 22:01 |
van51 | sonney2k: i'll look into it | 22:03 |
@sonney2k | van51, it is not your fault though -that thing was failing from time to time already | 22:06 |
@sonney2k | and look at the diff | 22:06 |
@sonney2k | http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1133/steps/test%20python_modular/logs/stdio | 22:06 |
@sonney2k | -w SGVector<float64> 2 ({-0.8028154615399042}{0.8573335463636752}) | 22:06 |
@sonney2k | +w SGVector<float64> 2 ({-0.8039184503697107}{0.8565511265518505}) | 22:06 |
@sonney2k | we have some *small* deviations | 22:06 |
@sonney2k | but dont' understand where these come from | 22:06 |
van51 | aah, I remember you discussing about those here | 22:07 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * eb9d67b / src/ (5 files): https://github.com/shogun-toolbox/shogun/commit/eb9d67bcaf8345bf1b199b528d447ffef618329d | 22:15 |
shogun-notifier- | shogun: define HAVE_SSE2 in lib/config.h too and fix a few doxygen errors | 22:15 |
shogun-buildbot | build #841 of cyg1 - libshogun is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/841 blamelist: Soeren Sonnenburg <sonne@debian.org> | 22:27 |
-!- travis-ci [~travis-ci@ec2-75-101-250-159.compute-1.amazonaws.com] has joined #shogun | 22:36 | |
travis-ci | [travis-ci] it's Soeren Sonnenburg's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/7284935 | 22:36 |
-!- travis-ci [~travis-ci@ec2-75-101-250-159.compute-1.amazonaws.com] has left #shogun [] | 22:36 | |
shogun-notifier- | shogun-data: Soeren Sonnenburg :master * 3fa1fdf / testsuite/tests/evaluation_multiclassovrevaluation_modular0.txt: https://github.com/shogun-toolbox/shogun-data/commit/3fa1fdfc6e0abb9bfaf7d0733ac34d5140914348 | 22:37 |
shogun-notifier- | shogun-data: another attempt to fix the test | 22:37 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * 5edcf8a / / (2 files): https://github.com/shogun-toolbox/shogun/commit/5edcf8a6945fd2d538e017c889bccb8db38fb1b5 | 22:38 |
shogun-notifier- | shogun: simplify example | 22:38 |
-!- van51 [~van51@athedsl-320452.home.otenet.gr] has left #shogun ["PING 1368910020"] | 22:47 | |
-!- hushell [~hushell@c-67-189-100-116.hsd1.or.comcast.net] has quit [Quit: leaving] | 22:56 | |
shogun-buildbot | build #842 of cyg1 - libshogun is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/842 | 22:56 |
-!- travis-ci [~travis-ci@ec2-107-22-88-191.compute-1.amazonaws.com] has joined #shogun | 22:59 | |
travis-ci | [travis-ci] it's Soeren Sonnenburg's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/7285539 | 22:59 |
-!- travis-ci [~travis-ci@ec2-107-22-88-191.compute-1.amazonaws.com] has left #shogun [] | 22:59 | |
-!- van51 [~van51@athedsl-320452.home.otenet.gr] has joined #shogun | 23:05 | |
shogun-buildbot | build #1134 of deb3 - modular_interfaces is complete: Failure [failed test python_modular] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1134 blamelist: Soeren Sonnenburg <sonne@debian.org> | 23:06 |
-!- hushell [~hushell@c-67-189-100-116.hsd1.or.comcast.net] has joined #shogun | 23:10 | |
@sonney2k | van51, my guess now is either some kind of randomness in the algorithm or multi-threading | 23:18 |
@sonney2k | but hmmh | 23:18 |
@sonney2k | hey hushell - how is your nips paper going? | 23:18 |
@sonney2k | hah | 23:19 |
@sonney2k | found it | 23:19 |
@sonney2k | int j = i+rand()%(active_size-i); | 23:19 |
van51 | sonney2k: good job :D | 23:20 |
hushell | hey sonney2k! I am finishing it there are few experiments need to be done | 23:21 |
hushell | I'll have 2 meetings with my supervisor to disucss the introduction and then I can move to gsoc :) | 23:22 |
@sonney2k | lisitsyn, look at this shogun/lib/tapkee/external/barnes_hut_sne/tsne.hpp:x = 2 * (rand() / ((double) RAND_MAX + 1)) - 1; | 23:22 |
@sonney2k | van51, we have lots of rand() still in there | 23:22 |
@lisitsyn | sonney2k: ugly indeed | 23:22 |
@sonney2k | liblinear, lots of tapkee, murmurhash ... | 23:23 |
hushell | the weechat is really nice. I can hang out forever here | 23:26 |
shogun-buildbot | build #1135 of deb3 - modular_interfaces is complete: Failure [failed test python_modular] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1135 blamelist: Soeren Sonnenburg <sonne@debian.org> | 23:29 |
@sonney2k | hushell, when is the deadline? next week? | 23:29 |
hushell | May 31 | 23:30 |
hushell | but I don't want to wait until last day | 23:30 |
@sonney2k | hushell, wow! good planning | 23:32 |
@sonney2k | we *never* finished early | 23:32 |
@sonney2k | all the nips papers I was co-author on we finished around 5 am or later :D | 23:32 |
@sonney2k | on the day of the deadline of course | 23:32 |
hushell | :D I hope I can finish early but who knows | 23:33 |
@sonney2k | lisitsyn, maybe you could make that part of the interface so it could call shogun's random? | 23:33 |
@sonney2k | hushell, yeah... | 23:33 |
@lisitsyn | sonney2k: yes sure | 23:33 |
@lisitsyn | once I am done with that shitty opencl assignment :D | 23:34 |
@sonney2k | lisitsyn, that also explains why test one unit test fails sometimes? | 23:34 |
@lisitsyn | sonney2k: unfortunately not | 23:34 |
@lisitsyn | isomap is not random | 23:34 |
@sonney2k | lisitsyn, also not multithreaded I guess? | 23:41 |
@lisitsyn | sonney2k: multithreaded | 23:41 |
@lisitsyn | but openmp | 23:41 |
@lisitsyn | so not in shogun | 23:41 |
@sonney2k | so it is multithreaded but disabled? | 23:43 |
@sonney2k | or is openmp on? | 23:43 |
@sonney2k | lisitsyn, I wonder why liblinear tests never failed | 23:45 |
@sonney2k | rand is also crappido | 23:45 |
@lisitsyn | sonney2k: yes, we don't have openmp enabled | 23:45 |
@sonney2k | lisitsyn, ok | 23:46 |
shogun-notifier- | shogun-data: Soeren Sonnenburg :master * c35bd0e / testsuite/tests/evaluation_multiclassovrevaluation_modular0.txt: https://github.com/shogun-toolbox/shogun-data/commit/c35bd0e9cd97ce65f5f11756b9655bcee9c91c61 | 23:51 |
shogun-notifier- | shogun-data: regenerate with non random liblinear | 23:51 |
@sonney2k | hmmhh wiking are you sure about CMath::random? | 23:55 |
@sonney2k | I have some concerns that this won't work | 23:55 |
@sonney2k | wiking, I take it back :D | 23:58 |
--- Log closed Sun May 19 00:00:08 2013 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!