--- Log opened Tue Dec 19 00:00:43 2017 | ||
-!- sanuj [~sanuj@106.51.25.170] has joined #shogun | 03:13 | |
-!- sanuj [~sanuj@106.51.25.170] has quit [Remote host closed the connection] | 03:30 | |
-!- sanuj [~sanuj@106.51.25.170] has joined #shogun | 03:31 | |
-!- sanuj [~sanuj@106.51.25.170] has quit [Ping timeout: 240 seconds] | 03:35 | |
-!- sanuj [~sanuj@106.51.25.170] has joined #shogun | 04:13 | |
-!- sanuj [~sanuj@106.51.25.170] has quit [Remote host closed the connection] | 04:18 | |
-!- sanuj [~sanuj@106.51.25.170] has joined #shogun | 04:20 | |
-!- sanuj [~sanuj@106.51.25.170] has quit [Remote host closed the connection] | 04:21 | |
-!- sanuj [~sanuj@106.51.25.170] has joined #shogun | 04:21 | |
-!- sanuj [~sanuj@106.51.25.170] has quit [Read error: Connection reset by peer] | 04:26 | |
-!- sanuj [~sanuj@106.51.25.170] has joined #shogun | 04:31 | |
-!- hersch [6741c40a@gateway/web/freenode/ip.103.65.196.10] has joined #shogun | 05:13 | |
-!- hersch [6741c40a@gateway/web/freenode/ip.103.65.196.10] has quit [Quit: Page closed] | 05:20 | |
-!- sanuj [~sanuj@106.51.25.170] has quit [Remote host closed the connection] | 06:44 | |
-!- sanuj [~sanuj@106.51.25.170] has joined #shogun | 06:45 | |
-!- sanuj [~sanuj@106.51.25.170] has quit [Ping timeout: 260 seconds] | 06:49 | |
-!- braceletboy [uid256084@gateway/web/irccloud.com/x-oqawopahtubbdagz] has joined #shogun | 07:02 | |
-!- sanuj [~sanuj@106.51.25.170] has joined #shogun | 07:10 | |
-!- sanuj [~sanuj@106.51.25.170] has quit [Remote host closed the connection] | 07:17 | |
-!- sanuj [~sanuj@223.237.229.83] has joined #shogun | 07:27 | |
-!- sanuj [~sanuj@223.237.229.83] has quit [Remote host closed the connection] | 07:39 | |
-!- sanuj_ [~sanuj@106.51.66.13] has joined #shogun | 07:46 | |
-!- sanuj_ [~sanuj@106.51.66.13] has quit [Remote host closed the connection] | 08:20 | |
-!- sanuj [~sanuj@106.51.66.13] has joined #shogun | 08:21 | |
-!- sanuj [~sanuj@106.51.66.13] has quit [Remote host closed the connection] | 08:22 | |
-!- sanuj [~sanuj@106.51.66.13] has joined #shogun | 08:22 | |
-!- sanuj [~sanuj@106.51.66.13] has quit [Remote host closed the connection] | 09:18 | |
-!- sanuj [~sanuj@106.51.66.13] has joined #shogun | 09:18 | |
-!- sanuj [~sanuj@106.51.66.13] has quit [Ping timeout: 248 seconds] | 09:23 | |
-!- sanuj [~sanuj@106.51.66.13] has joined #shogun | 09:42 | |
-!- sanuj [~sanuj@106.51.66.13] has quit [] | 09:47 | |
-!- sanuj [~sanuj@106.51.66.13] has joined #shogun | 09:53 | |
-!- stoic [2d40dc7a@gateway/web/freenode/ip.45.64.220.122] has quit [Quit: Page closed] | 10:30 | |
lisitsyn | wiking: pingyy | 10:32 |
---|---|---|
-!- sanuj [~sanuj@106.51.66.13] has quit [Remote host closed the connection] | 10:53 | |
-!- sanuj [~sanuj@106.51.66.13] has joined #shogun | 10:54 | |
@wiking | lisitsyn, pong | 10:57 |
lisitsyn | haha | 10:59 |
lisitsyn | ok now irc | 10:59 |
-!- HeikoS [~heiko@81.91.246.181] has joined #shogun | 11:42 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 11:42 | |
@wiking | sonney2k, https://tomassetti.me/introduction-to-webassembly/ | 11:51 |
@wiking | and for all the others as well HeikoS lisitsyn ^ | 11:51 |
@HeikoS | wiking: thx! | 11:51 |
@HeikoS | wiking: Ill set up a smart share for the $$$ | 11:51 |
@HeikoS | tonight or so | 11:51 |
@wiking | kk | 11:51 |
@wiking | HeikoS, r u on splitwise? | 11:56 |
@HeikoS | think | 11:56 |
@wiking | i guess u use your gmail account for such things :) | 11:57 |
-!- travis-ci [~travis-ci@ec2-54-87-45-54.compute-1.amazonaws.com] has joined #shogun | 12:01 | |
travis-ci | it's Sergey Lisitsyn's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/318526381 | 12:01 |
-!- travis-ci [~travis-ci@ec2-54-87-45-54.compute-1.amazonaws.com] has left #shogun [] | 12:01 | |
-!- HeikoS [~heiko@81.91.246.181] has quit [Ping timeout: 240 seconds] | 12:04 | |
@wiking | buhuhu | 12:05 |
@wiking | major changes causes 50+ mins compilation times on travis :) | 12:05 |
-!- HeikoS [~heiko@81.91.246.181] has joined #shogun | 12:28 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 12:28 | |
lisitsyn | HeikoS: I think I'd go with just one more type, non-owning-any | 12:37 |
lisitsyn | this way we can use std::any | 12:37 |
lisitsyn | and drop that additional one later | 12:37 |
@HeikoS | that is good! | 12:38 |
@HeikoS | yes then we can drop the code, and just have minimal own stuff | 12:38 |
lisitsyn | HeikoS: also, get/put is in | 12:38 |
@HeikoS | great! | 12:38 |
@HeikoS | let me know once the SG_ADD workds | 12:38 |
@HeikoS | then I can start doing the model selection updates and cereal | 12:38 |
lisitsyn | oh that's kind of revolution, you will know for sure | 12:38 |
@HeikoS | lisitsyn: until then, I can work on getters/setters ... | 12:38 |
@HeikoS | i.e. prepare the code to use tags | 12:39 |
lisitsyn | HeikoS: I want to merge this kernel thing | 12:39 |
lisitsyn | kernel_machine | 12:40 |
lisitsyn | HeikoS: I'd actually be happy if you add a few more 'create-by-name' things once this works | 12:41 |
lisitsyn | I mean it is easy and we can get at least this thing done | 12:41 |
lisitsyn | HeikoS: other tricky patch you can come up with is like detecting whether it could be created by name in meta examples | 12:41 |
lisitsyn | and then use SomeClass() if it is not supported yet and blabla('some_class') if it is | 12:42 |
@wiking | lisitsyn, btw | 12:44 |
@wiking | couldn't we template this :) | 12:44 |
lisitsyn | template what? | 12:45 |
@wiking | create_lalala | 12:45 |
lisitsyn | it is templated | 12:45 |
lisitsyn | create_object<T> | 12:45 |
@wiking | you do specialize it in c++ | 12:45 |
@wiking | ? | 12:45 |
lisitsyn | but you need to specialize some of them | 12:45 |
@wiking | because imho maybe it's not even necessary? | 12:45 |
@wiking | because you know in swig we need to anyway | 12:45 |
@wiking | instantiate the template | 12:45 |
@wiking | to be available for swig | 12:45 |
lisitsyn | yeah but I think it is better if you have the same function in C++ also | 12:46 |
@wiking | %template(lalalalCreate) create_object<LalaLand> | 12:46 |
lisitsyn | I mean otherwise in our C++ examples | 12:46 |
@wiking | but isn't this api actually mostly for the swig interface | 12:46 |
@wiking | ? | 12:46 |
lisitsyn | it would be create_object<CKernelMachine>("LibSVM") | 12:46 |
lisitsyn | and in swig it would be | 12:46 |
lisitsyn | kernel_machine("LibSVM") | 12:46 |
@wiking | yep | 12:46 |
lisitsyn | it would be symmetrical to have the same thing in C++ | 12:47 |
@wiking | i mean i see your point of having it unified | 12:47 |
@wiking | but it's really just syntactic sugar in c++ | 12:47 |
@wiking | anyhow | 12:47 |
@wiking | sure | 12:47 |
lisitsyn | yeah it is | 12:47 |
@wiking | just sayiung that we should use swig's feature | 12:47 |
@wiking | see meta_r_fix feature branch :P | 12:47 |
lisitsyn | ah we don't need it this way | 12:47 |
lisitsyn | I mean it already works without any swig code | 12:47 |
@wiking | ah fuck that needs to be merged before js | 12:47 |
lisitsyn | which is good because I don't like swig code | 12:47 |
lisitsyn | :D | 12:47 |
@wiking | lisitsyn, yeah i mean honestly | 12:48 |
@wiking | we dont use swig really in a smart way | 12:48 |
lisitsyn | yeah it has stuff | 12:48 |
@wiking | i bet none of us really ever read the manual :P | 12:48 |
@wiking | (based on the code) | 12:48 |
lisitsyn | I did a lot of times :P | 12:48 |
lisitsyn | especially about ref/unref things | 12:48 |
@wiking | :) | 12:48 |
@wiking | ok lemme first check buildbot | 12:48 |
@wiking | btw | 12:49 |
@wiking | can u guess why that malloc->calloc thing was necessary | 12:49 |
@wiking | for dotiterators? | 12:49 |
lisitsyn | that bugfix? | 12:49 |
@wiking | the whole change | 12:49 |
@wiking | why we need the mem to be inited | 12:49 |
@wiking | anyhow | 12:50 |
lisitsyn | where is it? | 12:50 |
@wiking | lemme get u the diff | 12:50 |
@wiking | https://github.com/shogun-toolbox/shogun/commit/6174f819362a179dd0d3f319a1458c32620f27b7 | 12:50 |
lisitsyn | what does it fix? | 12:51 |
@wiking | when we merged dotiterator | 12:51 |
@wiking | then basically some tests failed | 12:51 |
@wiking | hence this fix | 12:51 |
lisitsyn | not a good sign IMO | 12:51 |
@wiking | yeah | 12:52 |
lisitsyn | "Should fix DotIterator::add() failing unittest due to non initialized memory." | 12:52 |
lisitsyn | let me check | 12:52 |
-!- sukey [~nodebot@188.165.194.51] has joined #shogun | 12:52 | |
-!- mode/#shogun [+o sukey] by ChanServ | 12:52 | |
@wiking | sukey, flip | 12:53 |
@sukey | (????)?? ??? | 12:53 |
-!- travis-ci [~travis-ci@ec2-54-163-53-110.compute-1.amazonaws.com] has joined #shogun | 12:55 | |
travis-ci | it's Sergey Lisitsyn's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/318526381 | 12:55 |
-!- travis-ci [~travis-ci@ec2-54-163-53-110.compute-1.amazonaws.com] has left #shogun [] | 12:55 | |
-!- sanuj [~sanuj@106.51.66.13] has quit [Remote host closed the connection] | 13:06 | |
-!- esbensorig [~esben@gw00-nor.sagitta.dk] has joined #shogun | 13:06 | |
-!- esbensorig [~esben@gw00-nor.sagitta.dk] has left #shogun [] | 13:06 | |
-!- sanuj [~sanuj@106.51.66.13] has joined #shogun | 13:07 | |
-!- sorig [~esben@gw00-nor.sagitta.dk] has joined #shogun | 13:07 | |
-!- sorig_ [~sorig@gw00-nor.sagitta.dk] has joined #shogun | 13:10 | |
lisitsyn | wiking: hey is it a known bug that "make test" causes to run tests against installed python package? | 13:11 |
lisitsyn | PYTHONPATH=/usr/local/lib/python2.7/dist-packages: | 13:11 |
lisitsyn | that's my pypath when I run python tests | 13:12 |
-!- sanuj [~sanuj@106.51.66.13] has quit [Ping timeout: 264 seconds] | 13:12 | |
-!- sorig [~esben@gw00-nor.sagitta.dk] has quit [Ping timeout: 256 seconds] | 13:12 | |
-!- sorig_ is now known as sorig | 13:12 | |
lisitsyn | HeikoS: ^ | 13:13 |
@HeikoS | lisitsyn: ? | 13:13 |
@HeikoS | details? | 13:14 |
lisitsyn | HeikoS: meta runs against installed python package | 13:14 |
@HeikoS | yes | 13:14 |
lisitsyn | I have all of them crashing as I didn't install it | 13:14 |
@HeikoS | yes | 13:14 |
lisitsyn | known bug? | 13:14 |
@HeikoS | tests are executed against installed | 13:14 |
@HeikoS | I saw that before | 13:15 |
lisitsyn | we need to fix that | 13:15 |
lisitsyn | otherwise I just don't know if it works :D | 13:15 |
@HeikoS | you mean it should be executed against the build directory? | 13:15 |
lisitsyn | yes | 13:16 |
lisitsyn | makes no sense to test installed | 13:16 |
lisitsyn | HeikoS: it looks like tester.cpp is to be patched? | 13:17 |
@HeikoS | I think I agree | 13:17 |
@HeikoS | but I also remember that i discussed it with wiking | 13:17 |
lisitsyn | we should use basically nothing out of build I am pretty sure | 13:17 |
lisitsyn | because build is like a container | 13:18 |
lisitsyn | out of build -> other than build | 13:18 |
@wiking | lisitsyn, not easy :) | 13:20 |
@wiking | lisitsyn, some cases work some are not | 13:20 |
@wiking | imo python used to work | 13:20 |
@wiking | R does not | 13:20 |
-!- sanuj [~sanuj@106.51.66.13] has joined #shogun | 13:20 | |
lisitsyn | https://pastebin.com/MmmrWpPk | 13:21 |
lisitsyn | anyone knows how to fix that? | 13:21 |
-!- sanuj [~sanuj@106.51.66.13] has quit [Remote host closed the connection] | 13:22 | |
lisitsyn | install shogun? :D | 13:22 |
@wiking | mmm | 13:22 |
@wiking | that should be easy to fix | 13:22 |
@wiking | dunno when was this fucked | 13:22 |
@wiking | for python interface | 13:22 |
-!- sanuj [~sanuj@106.51.66.13] has joined #shogun | 13:22 | |
@wiking | you just need to be able to set the PYTHONPATH | 13:23 |
@HeikoS | thats how I solve it locally | 13:23 |
lisitsyn | well yeah | 13:23 |
@HeikoS | when I run tests | 13:23 |
@wiking | that used to be like set for the build/src/interface/python | 13:23 |
@wiking | but then i guess somebody changed it | 13:23 |
@wiking | i guess in meta examples | 13:23 |
@wiking | there's no if for this | 13:23 |
lisitsyn | ?\_(?)_/? | 13:23 |
@wiking | mmm | 13:24 |
@wiking | there is | 13:24 |
@wiking | then i dont get it | 13:24 |
lisitsyn | really? | 13:24 |
@wiking | lisitsyn, PythonEnvironment.cmake | 13:24 |
@wiking | 406105ec66 (Bj?rn Esser 2013-09-18 00:21:09 +0200 4) LIST(APPEND PYTHON_ENV_VARS PYTHONPATH=${CMAKE_INSTALL_PREFIX}/${PYTHON_PACKAGES_PATH}:$ENV{PYTHONPATH}) | 13:24 |
@wiking | :) | 13:24 |
@wiking | so bjorn changed this | 13:24 |
@wiking | in 406105ec66 | 13:25 |
lisitsyn | 2013? | 13:25 |
lisitsyn | ?? | 13:25 |
lisitsyn | :D | 13:25 |
@wiking | if u blame the rest of the file u see that i did it right :) | 13:25 |
lisitsyn | sorry | 13:25 |
@wiking | commit 406105ec66ca065314e068fa6d547eb6e7a08110 | 13:25 |
@wiking | Author: Bj?rn Esser <bjoern.esser@gmail.com> | 13:25 |
@wiking | Date: Wed Sep 18 00:21:09 2013 +0200 | 13:25 |
@wiking | thou shalt respect $ENV{PYTHONPATH} | 13:25 |
@wiking | :) | 13:25 |
lisitsyn | eh? | 13:25 |
lisitsyn | I guess this has something to do with fedora packaging | 13:25 |
@wiking | mmmm it should not | 13:25 |
@wiking | but anyhow | 13:25 |
@wiking | just send in a patch | 13:25 |
@wiking | where | 13:25 |
@wiking | PYTHONPATH=${CMAKE_BUILD_DIR}/src/shoguin/intrafaces/python | 13:26 |
@wiking | afaik | 13:26 |
lisitsyn | let me check if it works for me | 13:27 |
-!- sanuj [~sanuj@106.51.66.13] has quit [Ping timeout: 248 seconds] | 13:27 | |
@wiking | LIST(APPEND PYTHON_ENV_VARS PYTHONPATH=${CMAKE_BUILD_DIR}/src/shoguin/intrafaces/python:$ENV{PYTHONPATH}) | 13:27 |
@wiking | should be something like this | 13:27 |
@wiking | no | 13:27 |
@wiking | wrong :) | 13:27 |
@wiking | LIST(APPEND PYTHON_ENV_VARS PYTHONPATH=${CMAKE_BUILD_DIR}/src/intrafaces/python:$ENV{PYTHONPATH}) | 13:28 |
@wiking | :D:D:D | 13:28 |
@wiking | hahahah | 13:28 |
@wiking | ok i'm still tired | 13:28 |
@wiking | LIST(APPEND PYTHON_ENV_VARS PYTHONPATH=${CMAKE_BUILD_DIR}/src/interfaces/python:$ENV{PYTHONPATH}) | 13:28 |
@wiking | done | 13:28 |
@wiking | :) | 13:28 |
lisitsyn | eeh? | 13:28 |
@wiking | many typos there | 13:28 |
@wiking | last one should work | 13:28 |
@wiking | that should add the build dir of the python interface to the PYTHONPATH | 13:28 |
lisitsyn | I don't get why we have testing against installed | 13:28 |
lisitsyn | I mean it is intentional | 13:29 |
@wiking | imo it's because ASDF!@# | 13:29 |
lisitsyn | but it is not the proper way to do that for sure | 13:29 |
lisitsyn | I think it longs from pre-cmake times | 13:29 |
@wiking | maybe | 13:29 |
lisitsyn | because in cmake we have hermetic build | 13:29 |
@wiking | ignore | 13:29 |
@wiking | and do the right thing | 13:29 |
@wiking | just change it | 13:29 |
@wiking | test it | 13:30 |
@wiking | and pushit | 13:30 |
lisitsyn | ok I patched it locally, but we finally can go | 13:30 |
lisitsyn | budapest is waiting | 13:30 |
@wiking | :)))) | 13:30 |
@wiking | lisitsyn, u want me to push the fix? | 13:30 |
lisitsyn | wiking: I'll push once it fixes my problem | 13:31 |
@wiking | kk | 13:31 |
lisitsyn | PYTHONPATH=/src/interfaces/python:/usr/local/lib/python2.7/dist-packages: | 13:31 |
lisitsyn | MEEEH | 13:31 |
@wiking | asd? | 13:31 |
lisitsyn | wrong! | 13:31 |
@wiking | ahhahaha | 13:31 |
@wiking | CMAKE_BINARY_DIR | 13:31 |
@wiking | :D | 13:31 |
@wiking | LIST(APPEND PYTHON_ENV_VARS PYTHONPATH=${CMAKE_BINARY_DIR}/src/interfaces/python:$ENV{PYTHONPATH}) | 13:31 |
@wiking | dont listen to me today | 13:31 |
lisitsyn | CMAKE_BINARY! | 13:32 |
lisitsyn | :D | 13:32 |
@wiking | that's the lesson of today | 13:32 |
@wiking | ;))))) | 13:32 |
@wiking | 1 line and 4 typos | 13:32 |
@wiking | \o/ | 13:32 |
lisitsyn | yay | 13:32 |
@HeikoS | lisitsyn: let me know once the kernel is in and I will write a nicely detailed entrance task and make some example changes | 13:33 |
-!- sorig_ [~sorig@gw00-nor.sagitta.dk] has joined #shogun | 13:35 | |
@wiking | anybody knows what's the equivalent in R for is.numeric in case of bool? | 13:36 |
@wiking | is.logical | 13:36 |
@wiking | probbly | 13:36 |
@wiking | HeikoS, here? | 13:37 |
@HeikoS | wiking: not really, in working session | 13:37 |
@wiking | :D | 13:37 |
@HeikoS | I dont know the answer ;) | 13:37 |
@wiking | no it's not about that | 13:37 |
@wiking | shoulnd't we wanna use in case of R | 13:37 |
@wiking | the actual R datatypes | 13:37 |
@wiking | i'm thiking in case of meta_matrix | 13:38 |
@wiking | and meta_vector tests | 13:38 |
-!- sorig [~sorig@gw00-nor.sagitta.dk] has quit [Ping timeout: 268 seconds] | 13:38 | |
-!- sorig_ is now known as sorig | 13:38 | |
@wiking | and not the shogun ones | 13:38 |
@wiking | afaik in case of all the other languages | 13:38 |
@wiking | we use the native variable type | 13:38 |
@wiking | (python np.array, java doublematrix etc) | 13:38 |
@wiking | just in case of R we use RealVector, IntVector etc (shogun stuff) | 13:38 |
lisitsyn | 1/1 Test #700: generated_python-regression-krr_nystrom ...***Exception: SegFault 1.77 sec | 13:40 |
lisitsyn | yay much better | 13:40 |
@wiking | :) | 13:40 |
lisitsyn | uhm | 13:42 |
lisitsyn | ImportError: cannot import name WrappedObjectArray | 13:42 |
@HeikoS | ENABLE_TESTING=On | 13:43 |
@HeikoS | that thing is guarded | 13:43 |
@HeikoS | shouldnt happen this | 13:43 |
@HeikoS | so it is a bug | 13:43 |
@HeikoS | check the definition in the .h file | 13:43 |
@wiking | lisitsyn, cherry pick shit from the meta stuff | 13:43 |
@wiking | i dropped the whole Wrapped* thing | 13:43 |
@HeikoS | there is the guard, which is set by testing | 13:43 |
@wiking | lisitsyn, 76db3232c3c84957f6c1462d676b6fd09a3db5c0 | 13:44 |
-!- sorig [~sorig@gw00-nor.sagitta.dk] has quit [Ping timeout: 240 seconds] | 13:44 | |
lisitsyn | dropped? | 13:45 |
@wiking | yeah because that was redundant code | 13:45 |
@wiking | especially with type traits in c++11 | 13:45 |
@wiking | we now have Serializable | 13:45 |
@wiking | that should take care of shit | 13:46 |
lisitsyn | but it is in generated example?! | 13:46 |
lisitsyn | ahh | 13:46 |
lisitsyn | a branch | 13:46 |
@wiking | yes because of integration tests | 13:46 |
lisitsyn | it is in a branch right/ | 13:46 |
@wiking | yes | 13:46 |
@wiking | just rebased | 13:46 |
@wiking | over the latest develop | 13:46 |
@wiking | trying to fix the remaining errors in integration tests for R interface | 13:46 |
@wiking | and then i'll merge that branch | 13:46 |
@wiking | now only 11 integration tests are failing | 13:47 |
@wiking | instead of 52 | 13:47 |
@wiking | or no actually only 10 | 13:47 |
lisitsyn | nono I think that's rather what Heiko said | 13:47 |
lisitsyn | it was guarded away | 13:47 |
@wiking | yeah | 13:47 |
@wiking | but it will not be guarded anymore | 13:47 |
@wiking | if you just cherry it | 13:47 |
lisitsyn | super | 13:47 |
lisitsyn | I'd avoid :) | 13:48 |
@wiking | because there's no need to guard it | 13:48 |
-!- braceletboy [uid256084@gateway/web/irccloud.com/x-oqawopahtubbdagz] has quit [Quit: Connection closed for inactivity] | 13:48 | |
@wiking | it's just a class that adds serialization to basic types | 13:48 |
lisitsyn | yeah I know | 13:48 |
@wiking | okok | 13:48 |
lisitsyn | ok segfault was coming from LD | 13:48 |
@wiking | :> | 13:48 |
lisitsyn | we need the same thing for LD | 13:48 |
lisitsyn | like path | 13:48 |
@wiking | yeah there's the stuff | 13:48 |
@wiking | in that .cmake | 13:48 |
lisitsyn | yeah just below | 13:48 |
lisitsyn | phew fragile | 13:49 |
@wiking | it actually handles all the shiato | 13:49 |
@wiking | lisitsyn, use a var | 13:49 |
@wiking | :) | 13:49 |
lisitsyn | eh? | 13:49 |
@wiking | define a new cmake cvar | 13:49 |
@wiking | and set that path everywhere | 13:49 |
@wiking | :) | 13:49 |
lisitsyn | ah | 13:49 |
@wiking | BAZDMEG_PATH=${CMAKE_BINARY_DIR}/src/interfaces/python | 13:49 |
@wiking | and then use that everywhere | 13:49 |
@wiking | in that script | 13:49 |
@wiking | only 9 tests remain! | 13:50 |
@wiking | :))) | 13:50 |
@wiking | amaaaazing | 13:50 |
lisitsyn | ADSADSDas | 13:54 |
lisitsyn | WRAPPED | 13:54 |
@wiking | :> | 13:58 |
lisitsyn | wiking: ok your patch is really wanted because if you don't set flags properly it does not run | 13:59 |
@wiking | hehehe | 14:00 |
@wiking | yeah i dont wanna merge it until we dont have the R interface settled | 14:00 |
@wiking | :) | 14:00 |
@wiking | but i'm working on it | 14:01 |
@wiking | i can eliminate quite quickly 2 errors out of 9 | 14:01 |
@wiking | or maybe even 5 | 14:01 |
-!- travis-ci [~travis-ci@ec2-54-163-53-110.compute-1.amazonaws.com] has joined #shogun | 14:07 | |
travis-ci | it's Viktor Gal's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/318580063 | 14:07 |
-!- travis-ci [~travis-ci@ec2-54-163-53-110.compute-1.amazonaws.com] has left #shogun [] | 14:07 | |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4034 opened by lisitsyn | 14:08 |
lisitsyn | wiking: ^ here is that patch | 14:08 |
@wiking | noup | 14:09 |
@wiking | or yes?:) | 14:09 |
@wiking | yes | 14:09 |
@wiking | :) | 14:09 |
@wiking | good | 14:09 |
@wiking | :. | 14:09 |
@wiking | :> | 14:09 |
@wiking | lets see what travis python test says and then we merge | 14:09 |
lisitsyn | yeah I reckon it breaks something | 14:09 |
lisitsyn | :D | 14:09 |
@wiking | lisitsyn, ok | 14:18 |
@wiking | https://travis-ci.org/shogun-toolbox/shogun/jobs/318610043 | 14:18 |
@wiking | green | 14:18 |
@wiking | merge it! | 14:19 |
@wiking | arg | 14:21 |
@wiking | anybody has an idea that in case of the meta example | 14:21 |
@wiking | linear_svm | 14:21 |
@wiking | why would i get for the bias in R TRUE? | 14:21 |
@wiking | i.e. b <- svm$get_bias() | 14:21 |
@wiking | b = TRUE | 14:21 |
@wiking | is the bias exactly 1.0? | 14:21 |
lisitsyn | it is biased | 14:21 |
lisitsyn | :D :D :D | 14:21 |
@wiking | or wtf :) | 14:22 |
@wiking | although still doesn't make sense | 14:22 |
@wiking | as the return value of get_bias is float64_t | 14:22 |
@wiking | so i dont get why R would map that to bool | 14:22 |
lisitsyn | some clash? | 14:22 |
@wiking | mmm noup | 14:23 |
-!- HeikoS [~heiko@81.91.246.181] has quit [Quit: Leaving.] | 14:23 | |
@wiking | SWIGEXPORT SEXP | 14:23 |
@wiking | R_swig_LinearMachine_get_bias ( SEXP self, SEXP s_swig_copy) | 14:23 |
@wiking | SEXP = float | 14:23 |
@wiking | afaik | 14:23 |
@wiking | lisitsyn, MERGE IT | 14:24 |
@wiking | ah no | 14:24 |
@wiking | it's some special pointer var :D | 14:24 |
lisitsyn | wiking: the last job to finish | 14:26 |
@wiking | doesn't matter | 14:26 |
@wiking | u only changed PYTHON shit | 14:27 |
@wiking | that job is done | 14:27 |
@wiking | merg it | 14:27 |
@wiking | :) | 14:27 |
lisitsyn | done | 14:27 |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4034 merged by lisitsyn | 14:27 |
@sukey | [https://github.com/shogun-toolbox/shogun] New commit https://github.com/shogun-toolbox/shogun/commit/7132a0b6f9a14c975455a60cb480fcbc3915356d by lisitsyn | 14:27 |
@wiking | \o/ | 14:27 |
@wiking | perfect | 14:27 |
@wiking | lisitsyn, | 14:28 |
@wiking | super interesting shiatz | 14:29 |
@wiking | > svm$set_bias(2.1) | 14:29 |
@wiking | NULL | 14:29 |
@wiking | > b <- svm$get_bias() | 14:29 |
@wiking | > b | 14:29 |
@wiking | [1] TRUE | 14:29 |
@wiking | > | 14:29 |
@wiking | :))))) | 14:29 |
lisitsyn | uhm | 14:29 |
-!- sanuj [~sanuj@106.51.66.13] has joined #shogun | 14:29 | |
@wiking | > svm$set_w(c(1.0, 1.0)) | 14:29 |
@wiking | NULL | 14:29 |
@wiking | > svm$get_w() | 14:29 |
@wiking | [1] 1 1 | 14:29 |
@wiking | works | 14:29 |
lisitsyn | ok ttyl | 14:30 |
@wiking | ttyl | 14:31 |
@wiking | btw are we aware of the fact that CLeastAngleRegression is not serializable - in a sense that once trained and serialized it's not gonna get back the same obj? | 14:36 |
-!- travis-ci [~travis-ci@ec2-54-163-53-110.compute-1.amazonaws.com] has joined #shogun | 14:37 | |
travis-ci | it's Viktor Gal's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/318580063 | 14:37 |
-!- travis-ci [~travis-ci@ec2-54-163-53-110.compute-1.amazonaws.com] has left #shogun [] | 14:37 | |
-!- HeikoS [~heiko@81.91.246.181] has joined #shogun | 15:05 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 15:05 | |
-!- HeikoS [~heiko@81.91.246.181] has quit [Ping timeout: 248 seconds] | 15:20 | |
-!- sanuj [~sanuj@106.51.66.13] has quit [Remote host closed the connection] | 15:25 | |
-!- sanuj [~sanuj@106.51.66.13] has joined #shogun | 15:26 | |
-!- sanuj [~sanuj@106.51.66.13] has quit [Ping timeout: 260 seconds] | 15:31 | |
-!- sanuj [~sanuj@106.51.66.13] has joined #shogun | 15:39 | |
-!- HeikoS [~heiko@81.91.246.181] has joined #shogun | 15:49 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 15:49 | |
@wiking | HeikoS, i'll fwd u an email now :D | 16:02 |
-!- sanuj [~sanuj@106.51.66.13] has quit [Remote host closed the connection] | 16:09 | |
-!- sanuj [~sanuj@106.51.66.13] has joined #shogun | 16:12 | |
-!- sanuj [~sanuj@106.51.66.13] has quit [Remote host closed the connection] | 16:33 | |
-!- sanuj [~sanuj@106.51.66.13] has joined #shogun | 16:34 | |
-!- sanuj [~sanuj@106.51.66.13] has quit [Ping timeout: 272 seconds] | 16:39 | |
-!- sanuj [~sanuj@106.51.25.170] has joined #shogun | 17:08 | |
-!- sanuj [~sanuj@106.51.25.170] has quit [Ping timeout: 265 seconds] | 17:13 | |
-!- sanuj [~sanuj@106.51.25.170] has joined #shogun | 18:03 | |
-!- HeikoS [~heiko@81.91.246.181] has quit [Quit: Leaving.] | 18:20 | |
-!- HeikoS [~heiko@81.91.246.181] has joined #shogun | 18:20 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 18:20 | |
-!- HeikoS [~heiko@81.91.246.181] has quit [Ping timeout: 272 seconds] | 18:25 | |
-!- sanuj [~sanuj@106.51.25.170] has quit [Read error: Connection reset by peer] | 20:02 | |
-!- sanuj [~sanuj@106.51.25.170] has joined #shogun | 20:02 | |
-!- sanuj_ [~sanuj@106.51.25.170] has joined #shogun | 20:07 | |
-!- sanuj [~sanuj@106.51.25.170] has quit [Read error: Connection reset by peer] | 20:07 | |
-!- sanuj_ [~sanuj@106.51.25.170] has quit [Remote host closed the connection] | 20:14 | |
-!- sanuj [~sanuj@106.51.25.170] has joined #shogun | 20:32 | |
-!- sanuj [~sanuj@106.51.25.170] has quit [Remote host closed the connection] | 20:33 | |
-!- sanuj [~sanuj@106.51.25.170] has joined #shogun | 20:33 | |
-!- sanuj [~sanuj@106.51.25.170] has quit [Remote host closed the connection] | 20:34 | |
-!- sanuj [~sanuj@106.51.25.170] has joined #shogun | 20:34 | |
-!- sanuj [~sanuj@106.51.25.170] has quit [Remote host closed the connection] | 20:36 | |
-!- sanuj [~sanuj@106.51.25.170] has joined #shogun | 20:36 | |
-!- sanuj [~sanuj@106.51.25.170] has quit [Remote host closed the connection] | 20:38 | |
-!- sanuj [~sanuj@106.51.25.170] has joined #shogun | 20:38 | |
-!- sanuj_ [~sanuj@106.51.25.170] has joined #shogun | 20:53 | |
-!- sanuj [~sanuj@106.51.25.170] has quit [Ping timeout: 240 seconds] | 20:56 | |
-!- sanuj_ [~sanuj@106.51.25.170] has quit [Remote host closed the connection] | 21:11 | |
-!- sanuj [~sanuj@106.51.25.170] has joined #shogun | 21:32 | |
-!- sanuj [~sanuj@106.51.25.170] has quit [Ping timeout: 260 seconds] | 21:36 | |
-!- sanuj [~sanuj@106.51.25.170] has joined #shogun | 22:04 | |
-!- sanuj [~sanuj@106.51.25.170] has quit [Ping timeout: 264 seconds] | 22:09 | |
-!- HeikoS [~heiko@host86-171-158-219.range86-171.btcentralplus.com] has joined #shogun | 22:15 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 22:15 | |
-!- sanuj [~sanuj@106.51.25.170] has joined #shogun | 22:25 | |
lisitsyn | HeikoS: hey | 22:25 |
@HeikoS | lisitsyn: jojo | 22:26 |
lisitsyn | HeikoS: I want to discuss two things a bit later | 22:26 |
lisitsyn | 1) shogun kernel prize | 22:26 |
lisitsyn | 2) mloss thing | 22:26 |
lisitsyn | shogun kernel prize = kaggle | 22:26 |
@HeikoS | not sure what you mean by 1)? | 22:27 |
lisitsyn | I don't mind having a call, when is good for you? | 22:27 |
lisitsyn | HeikoS: we can set up a prize for kernels (notebooks) at kaggle | 22:27 |
lisitsyn | that use shogun | 22:28 |
lisitsyn | that's my idea I want to discuss | 22:28 |
lisitsyn | HeikoS: they have things like 'best kernel of week' | 22:28 |
-!- sanuj [~sanuj@106.51.25.170] has quit [Ping timeout: 272 seconds] | 22:30 | |
@HeikoS | lisitsyn: you we sponsor that? | 22:30 |
@HeikoS | best notebook that uses shogun ? | 22:30 |
@HeikoS | and we pick and give money to best one? | 22:30 |
@wiking | HeikoS, not all k is about $ | 22:30 |
lisitsyn | HeikoS: yes | 22:31 |
lisitsyn | wiking: oh you're here ;) | 22:31 |
@wiking | oh yes im here | 22:32 |
* wiking feels invisible | 22:32 | |
@wiking | :D | 22:32 |
lisitsyn | wiking: I have this thing: | 22:32 |
lisitsyn | The following tests FAILED: | 22:32 |
lisitsyn | 615 - generated_cpp-binary_classifier-mkl (OTHER_FAULT) | 22:32 |
lisitsyn | 632 - generated_cpp-meta_api-calls (SEGFAULT) | 22:32 |
lisitsyn | 662 - generated_python-binary_classifier-mkl (Failed) | 22:32 |
@wiking | hehe | 22:32 |
lisitsyn | 632 I want to debug | 22:32 |
@wiking | i have some of those as well | 22:33 |
lisitsyn | but others are just like unsupported | 22:33 |
lisitsyn | we need to guard them somehow | 22:33 |
lisitsyn | it is pretty bad 632 fails | 22:33 |
@wiking | those are some global calls | 22:33 |
@wiking | or one of the global calls fails | 22:33 |
lisitsyn | gdb says some bs | 22:34 |
@wiking | https://www.kaggle.com/c/jigsaw-toxic-comment-classification-challenge?utm_medium=email&utm_source=intercom&utm_campaign=jigsaw+competition+2017\ | 22:34 |
@wiking | just got out | 22:34 |
@wiking | :) | 22:34 |
@wiking | In this competition, you're challenged to build a multi-headed model that's capable of detecting different types of of toxicity like threats, obscenity, insults, and identity-based hate better than Perspective's current models. You'll be using a dataset of comments from Wikipedia's talk page edits. Improvements to the current model will hopefully help online discussion become more productive and respectful. | 22:34 |
@HeikoS | lisitsyn: so what is it that you want to do? | 22:35 |
@wiking | so nlp | 22:35 |
lisitsyn | HeikoS: ?? | 22:35 |
@HeikoS | with the kernel kaggle | 22:35 |
lisitsyn | :D | 22:35 |
lisitsyn | I want to discuss that idea but probably I just need to explain it here :D | 22:35 |
lisitsyn | so we get some money to sponsor this prize | 22:36 |
lisitsyn | and talk to kaggle people if we can do that | 22:36 |
lisitsyn | basically that's it | 22:36 |
@wiking | lisitsyn, donnt we just wanna do it for | 22:36 |
@wiking | 'learn' | 22:36 |
@wiking | :) | 22:36 |
lisitsyn | ? | 22:36 |
lisitsyn | don't get it | 22:36 |
@wiking | 'knowledge' | 22:36 |
lisitsyn | ah | 22:37 |
@wiking | there are knowledge competitions | 22:37 |
@wiking | that are not $ | 22:37 |
lisitsyn | no no | 22:37 |
lisitsyn | I am not talking about competition | 22:37 |
@wiking | see | 22:37 |
@wiking | https://www.kaggle.com/c/street-view-getting-started-with-julia | 22:37 |
@wiking | :) | 22:37 |
lisitsyn | I am talking about kernels | 22:37 |
@wiking | we could definitely do something similar | 22:37 |
lisitsyn | kernels are much better | 22:37 |
lisitsyn | they are inter-competition | 22:37 |
@HeikoS | lisitsyn: sorry can you ellaborate a bit more? | 22:37 |
lisitsyn | oh | 22:37 |
@HeikoS | like in: what is exactly the task? | 22:37 |
@HeikoS | or is there a website with infos? | 22:37 |
lisitsyn | ok so do you know what's the kaggle kernel? | 22:37 |
lisitsyn | basically jupyter notebook doing some code right? :) | 22:38 |
@wiking | lisitsyn, yeah i mean we can do that as well for one of the comp | 22:38 |
lisitsyn | and they award best kernels of week with some small $ | 22:38 |
@wiking | u wanna add extra $ for that kernel? | 22:38 |
lisitsyn | I suggest adding one more 'award' | 22:38 |
lisitsyn | best shogun kernel | 22:38 |
lisitsyn | say every month | 22:38 |
@HeikoS | I see | 22:38 |
lisitsyn | which is competition agnostic | 22:38 |
lisitsyn | just the best kernel that uses shogun | 22:38 |
@HeikoS | so then some students come and do that | 22:39 |
lisitsyn | yes | 22:39 |
@HeikoS | and who picks the best one? | 22:39 |
@HeikoS | we? | 22:39 |
@HeikoS | or voting? | 22:39 |
lisitsyn | uhm we'd have to talk to kaggle people I don't know | 22:39 |
lisitsyn | maybe its voting | 22:39 |
lisitsyn | voting is even better | 22:39 |
@HeikoS | less work | 22:39 |
@HeikoS | and why would they sponsor that? | 22:40 |
@HeikoS | what is our argument? | 22:40 |
@HeikoS | seems like some work | 22:40 |
lisitsyn | uhm I think that's we who sponsor that | 22:40 |
lisitsyn | :D | 22:40 |
lisitsyn | HeikoS: we can generalize this idea a bit to leverage numfocus | 22:41 |
lisitsyn | we can say we want to support some underused projects of numfocus | 22:41 |
lisitsyn | so we establish this monthly award to make people use pymc3 shogun and julia | 22:41 |
lisitsyn | :D | 22:41 |
lisitsyn | or whatever software we define not-that-used | 22:42 |
@wiking | COBOL | 22:42 |
@wiking | FORTRAN | 22:42 |
lisitsyn | yeah cobol kernel of week | 22:42 |
@wiking | ;) | 22:42 |
lisitsyn | HeikoS: so like/dislike? | 22:43 |
@HeikoS | lisitsyn: I like that idea a lot | 22:44 |
@HeikoS | especially with numdocus | 22:44 |
@HeikoS | we might even get them to sponsor a bit | 22:44 |
lisitsyn | yeah that's the point | 22:44 |
@HeikoS | I will write a proposal for that | 22:44 |
lisitsyn | can I help moving this thing? | 22:44 |
@HeikoS | and then we can re-cycle it for numfoczus/kaggle etc | 22:44 |
@HeikoS | yes, start a googledoc and put infos down | 22:44 |
@HeikoS | Then I can turn that into text | 22:44 |
@HeikoS | and then give it to lea to make it awesome ;) | 22:45 |
lisitsyn | community manager! | 22:45 |
lisitsyn | HeikoS: I think we need some kind of software to track such activities | 22:46 |
lisitsyn | can someone research on it a bit? | 22:46 |
lisitsyn | basecamp? trello? | 22:47 |
@wiking | trello | 22:47 |
@wiking | only | 22:47 |
@wiking | :) | 22:47 |
lisitsyn | yeah I don't care :) | 22:47 |
@HeikoS | no just the doc | 22:47 |
lisitsyn | HeikoS: yes for that thing | 22:47 |
@wiking | :D | 22:47 |
@HeikoS | will be enough | 22:47 |
lisitsyn | but in general | 22:47 |
lisitsyn | we need to track such activities | 22:47 |
@HeikoS | yep, but we also need to use it and that never happens :) | 22:48 |
@wiking | lisitsyn, HeikoS doesn't use trello | 22:48 |
@wiking | :) | 22:48 |
@HeikoS | github project? | 22:48 |
@HeikoS | I use trello every day actually | 22:48 |
@HeikoS | but I dont like another site for us, too many things to check | 22:48 |
@wiking | HeikoS, yeah hence your comment that you dont wanna use it | 22:48 |
@HeikoS | github project is fine | 22:48 |
@wiking | cca 1.5 years ago | 22:48 |
lisitsyn | I don't like github project as some things are not public | 22:48 |
lisitsyn | ah we can make it private sorry | 22:48 |
lisitsyn | ok I don't like github project because it is kind of immature | 22:49 |
lisitsyn | :D | 22:49 |
lisitsyn | ok it is up to you but I think we lack it | 22:49 |
@HeikoS | lol | 22:49 |
@HeikoS | I mean make a trello board | 22:49 |
@HeikoS | I will use it | 22:49 |
@HeikoS | But I would prefer to just do the article, ping each other every now and then | 22:49 |
@wiking | btw what do we do with the actions | 22:49 |
@HeikoS | and then have SG_ADD ready soon ;) | 22:49 |
@wiking | in the https://github.com/shogun-toolbox/shogun/projects/8 | 22:49 |
@HeikoS | populate the sites | 22:49 |
@HeikoS | I'm on it | 22:50 |
@HeikoS | but the list is long ... sadly | 22:50 |
@HeikoS | can send a reminder | 22:50 |
@HeikoS | to the attendees | 22:50 |
lisitsyn | sorry but we need some management :D | 22:50 |
@wiking | populate the sites? | 22:50 |
@wiking | what sites? | 22:50 |
@wiking | lisitsyn, https://chainer.org/ | 22:50 |
lisitsyn | yeah know this thing | 22:51 |
@HeikoS | lisitsyn: do it | 22:51 |
lisitsyn | wiking: not that different from any other | 22:51 |
@HeikoS | manage me | 22:52 |
@HeikoS | I am up for using it all | 22:52 |
lisitsyn | HeikoS: I didn't say I can do that :P | 22:52 |
lisitsyn | "SG_ADD" and setting tasks is way out of my capacity | 22:53 |
@wiking | lol https://www.johndcook.com/blog/2010/04/29/simple-approximation-to-normal-distribution/ | 22:54 |
@HeikoS | ok just SG_ADD! | 22:55 |
lisitsyn | HeikoS: yeah I just want to point we have too many places to check :) | 22:56 |
lisitsyn | *someone* could fix it | 22:56 |
@HeikoS | I will set up a trello board | 22:57 |
@HeikoS | good? | 22:57 |
@HeikoS | Ok I gotta eat/run now | 22:57 |
lisitsyn | HeikoS: if you think this solves :) | 22:57 |
@HeikoS | send the me doc link | 22:57 |
lisitsyn | okay | 22:57 |
@HeikoS | lisitsyn: no i dont think that, think we need an organiser | 22:57 |
@HeikoS | somebody who doesnt do, but just tells others what to do ;) | 22:57 |
@HeikoS | would love to be that person, but then there is even less people doing stuff | 22:57 |
@HeikoS | sleep well! | 22:57 |
@wiking | https://www.youtube.com/watch?v=ZXsQAXx_ao0 | 22:57 |
@wiking | Shia LaBeouf will tell us | 22:58 |
lisitsyn | ah that's good | 22:58 |
lisitsyn | shia driven is fine | 22:59 |
lisitsyn | :D | 22:59 |
@wiking | "if you are tired starting over, stop giving up | 22:59 |
@wiking | :D | 22:59 |
@HeikoS | hahaha | 23:00 |
lisitsyn | wiking: I am surprised someone does string kernels still | 23:00 |
@HeikoS | lisitsyn: actually I really like this idea | 23:00 |
@HeikoS | with the kaggle | 23:00 |
@wiking | lisitsyn, yeah dunno | 23:00 |
@wiking | lisitsyn, i mean the comment is the most interesting | 23:00 |
@wiking | 'super hard' | 23:00 |
lisitsyn | I was 100% sure everybody just convolute ACTG nowadays | 23:00 |
lisitsyn | HeikoS: it will be in doc so then the ball is on your side :) | 23:01 |
@HeikoS | good! | 23:01 |
lisitsyn | wiking: that's fine, shogun is really super hard :D | 23:01 |
lisitsyn | wiking: I am really curious if string kernels are still the way to go | 23:02 |
-!- HeikoS [~heiko@host86-171-158-219.range86-171.btcentralplus.com] has quit [Quit: Leaving.] | 23:02 | |
lisitsyn | is it like STACK MORE LAYERS does not work? | 23:02 |
lisitsyn | convolve these letters and it works, no? | 23:02 |
lisitsyn | :P | 23:03 |
lisitsyn | max pooling stride 3 whatever | 23:03 |
@wiking | STACK MORE LAYERS1 | 23:03 |
--- Log closed Wed Dec 20 00:00:44 2017 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!