--- Log opened Mon Mar 03 00:00:22 2014 | ||
-!- HeikoS [~heiko@2.219.252.217] has joined #shogun | 00:11 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 00:11 | |
-!- HeikoS [~heiko@2.219.252.217] has quit [Quit: Leaving.] | 01:10 | |
-!- Calvin-O [~mohit@c-24-118-110-65.hsd1.mn.comcast.net] has quit [Quit: Leaving.] | 01:27 | |
-!- HeikoS [~heiko@2.219.252.217] has joined #shogun | 01:27 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 01:27 | |
-!- HeikoS [~heiko@2.219.252.217] has quit [Ping timeout: 264 seconds] | 02:17 | |
-!- zxtx_ is now known as zxtx | 02:43 | |
-!- vaishaks [~vaishak@1.186.131.85] has joined #shogun | 03:06 | |
shogun-buildbot | build #632 of nightly_all is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/nightly_all/builds/632 | 03:08 |
---|---|---|
-!- kislay [~kislay@93.114.45.13] has joined #shogun | 04:52 | |
-!- lisitsyn [~lisitsyn@83-69-213-138.in-addr.mastertelecom.ru] has left #shogun [] | 05:40 | |
-!- sonne|osx [~sonne@89.204.135.203] has joined #shogun | 08:09 | |
-!- kislay [~kislay@93.114.45.13] has quit [Read error: Connection reset by peer] | 08:21 | |
-!- sonne|osx [~sonne@89.204.135.203] has quit [Ping timeout: 264 seconds] | 08:24 | |
-!- kislay [~kislay@93.114.45.13] has joined #shogun | 09:00 | |
-!- vaishaks [~vaishak@1.186.131.85] has quit [Ping timeout: 252 seconds] | 09:09 | |
-!- vaishaks [~vaishak@1.186.131.85] has joined #shogun | 09:09 | |
-!- atin [~mathurati@122.161.88.2] has joined #shogun | 09:11 | |
atin | hi | 09:11 |
@besser82 | atin: hi! | 09:12 |
-!- dhruv13 [~dhruv13@triband-mum-120.61.113.206.mtnl.net.in] has joined #shogun | 09:15 | |
-!- atin [~mathurati@122.161.88.2] has quit [] | 09:19 | |
dhruv13 | I need some help, | 09:20 |
dhruv13 | I created a function which does float comparisons, and which is working fine | 09:21 |
dhruv13 | but some json serialization test cases are failing | 09:21 |
dhruv13 | since they compare the deserialized object with the serialized one | 09:22 |
sonne|work | dhruv13: json doesn't have full precision | 09:22 |
dhruv13 | some floating point parameters initialized with really small values, like LMNN::stepsize = 1E-07 are truncated to 0 after deserialization | 09:23 |
dhruv13 | yeah | 09:23 |
dhruv13 | which is why the test cases fail | 09:23 |
dhruv13 | the fequals function seems to detect this for these parameters | 09:23 |
dhruv13 | the previous method for float comparison was able to tolerate this... | 09:23 |
dhruv13 | I'm not sure what to do | 09:24 |
dhruv13 | one solution is to introduce a relative vs. absolute epsilon tolerance for the comparison | 09:24 |
dhruv13 | but i think it's overkill | 09:25 |
dhruv13 | maybe edit test cases for JSON so that they use a different (error tolerant) way to compare the objects? | 09:26 |
dhruv13 | refer - https://github.com/shogun-toolbox/shogun/pull/1896 for the changes I made... | 09:27 |
dhruv13 | any suggestions? | 09:30 |
sonne|work | dhruv13: yes json needs lower precision in the tests - so that would be a good fix | 09:33 |
dhruv13 | hmm, i will change the function to accept a "tolerant" parameter | 09:34 |
dhruv13 | bool | 09:35 |
dhruv13 | which, if true, checks only against epsilon | 09:35 |
dhruv13 | thanks!! | 09:35 |
-!- kislay [~kislay@93.114.45.13] has quit [Read error: Connection reset by peer] | 09:42 | |
-!- kislay [~kislay@93.114.45.13] has joined #shogun | 09:57 | |
-!- kislay [~kislay@93.114.45.13] has quit [Read error: Connection reset by peer] | 10:02 | |
-!- bluenotebook [~bluenoteb@115.249.18.25] has joined #shogun | 10:08 | |
-!- vaishaks [~vaishak@1.186.131.85] has quit [Ping timeout: 240 seconds] | 10:17 | |
-!- vaishaks [~vaishak@1.186.131.85] has joined #shogun | 10:18 | |
bluenotebook | anyone who can help me with installing shogun? | 10:21 |
bluenotebook | on linus | 10:21 |
-!- aroma [01261213@gateway/web/freenode/ip.1.38.18.19] has joined #shogun | 10:21 | |
bluenotebook | linux* | 10:21 |
aroma | hi. i am having issues in building shogun-develop in ubuntu 13.04 . | 10:26 |
bluenotebook | ^ditto | 10:28 |
-!- vaishaks [~vaishak@1.186.131.85] has quit [Ping timeout: 264 seconds] | 10:29 | |
-!- vaishaks [~vaishak@1.186.131.85] has joined #shogun | 10:30 | |
sonne|work | bluenotebook: aromawell then what are they :D | 10:31 |
sonne|work | might be worth to share the secret! | 10:31 |
-!- aroma [01261213@gateway/web/freenode/ip.1.38.18.19] has quit [Ping timeout: 245 seconds] | 10:32 | |
-!- aroma [01261405@gateway/web/freenode/ip.1.38.20.5] has joined #shogun | 10:33 | |
aroma | although it says atlas and lapack are optional | 10:35 |
aroma | it giyes compliation error when atlas and lapack are not installed | 10:35 |
aroma | exact error message is : | 10:37 |
aroma | make[2]: *** [src/shogun/CMakeFiles/libshogun.dir/ui/GUIPreprocessor.cpp.o] Error 1 make[1]: *** [src/shogun/CMakeFiles/libshogun.dir/all] Error 2 make: *** [all] Error 2 | 10:37 |
aroma | i looked into this file . the error is in part of the code which uses lapack | 10:40 |
bluenotebook | I think my error is solved. Apologies for bothering you. :) | 10:41 |
-!- vaishaks [~vaishak@1.186.131.85] has quit [Ping timeout: 240 seconds] | 10:41 | |
-!- vaishaks [~vaishak@1.186.131.85] has joined #shogun | 10:42 | |
-!- tom_ [2eda6d58@gateway/web/freenode/ip.46.218.109.88] has joined #shogun | 10:43 | |
tom_ | hi all! | 10:43 |
-!- tom_ is now known as Guest49407 | 10:43 | |
-!- aroma [01261405@gateway/web/freenode/ip.1.38.20.5] has quit [Ping timeout: 245 seconds] | 10:44 | |
Guest49407 | I would like some advice in using SG_REF and SG_UNREF: the cpp examples show different usage convention .. | 10:44 |
Guest49407 | should I always SG_REF a new *obj ? | 10:45 |
sonne|work | bluenotebook: well enlighten us | 10:45 |
sonne|work | aroma that is a bug then | 10:46 |
bluenotebook | There was some problem with importing from shogun in python | 10:47 |
bluenotebook | I couldn't run the example programs for python modular | 10:48 |
bluenotebook | I thought shogun didn't install properly. I re-installed and voila | 10:48 |
-!- aroma [0e8b2998@gateway/web/freenode/ip.14.139.41.152] has joined #shogun | 10:48 | |
bluenotebook | everything seems to be working | 10:48 |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 10:49 | |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * ad04382 / src/shogun/ui/GUIPreprocessor.cpp: https://github.com/shogun-toolbox/shogun/commit/ad043824822ef95b9882a5fa54806d0d67f6e2ee | 10:49 |
shogun-notifier- | shogun: fix compile error when building w/o eigen3 | 10:49 |
lambday | Guest49407: that depends on your application.. | 10:52 |
bluenotebook | Also, Sonney, I am still working on making a meta-language for shogun examples. I have my mid-terms starting tomorrow, hence the delay. | 10:52 |
bluenotebook | I am running too late? | 10:53 |
lambday | Guest49407: SG_REF thing is used for reference counting.. so if you create an object somewhere and use it somewhere else (say, inside another object), and that object gets destroyed... you're left with a dangling ptr... | 10:53 |
lambday | Guest49407: its the same thing that c++11 smart ptrs do - here its done explicitly | 10:55 |
lambday | Guest49407: the common convention is, if you have a class that takes another class objs in its constructor, we usually SG_REF it - and UNREF it in the destructor | 10:56 |
lambday | just to make sure that thing lives in memory as long as we wished it to | 10:57 |
sonne|work | bluenotebook: well decision about things is way in the future... so no | 10:58 |
bluenotebook | Alright, thanks. I should come up with something substantial till D day :) | 10:58 |
bluenotebook | Much before it * | 10:59 |
sonne|work | bluenotebook: *I* would do it just with jinja2 templates and very simple syntax | 10:59 |
sonne|work | Guest49407, lambday - yeah you only have to SG_REF if you pass the object somewhere since inside such a class / function it might be unref'd and then your local variable might be invalid. | 11:00 |
lambday | sonne|work: yeah... | 11:00 |
lambday | sonne|work: btw should we replace it with shared_ptr some pt of time? since c++11 support is already there - would be a lot of work though | 11:01 |
sonne|work | lambday: yes would be nice - basically same thing like we do with the SGVector etc auto refcounting stuff | 11:02 |
-!- HeikoS [~heiko@2.219.252.217] has joined #shogun | 11:02 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 11:02 | |
sonne|work | lambday: I am afraid that this is something to be done after gsoc though not during :) | 11:04 |
lambday | sonne|work: lol yeah! :D | 11:04 |
lambday | HeikoS: hi! I slept early last night... trying to figure out this data thing | 11:04 |
sonne|work | bluenotebook: well I would start with trying to get all interfaces to compile | 11:05 |
sonne|work | then take e.g. the libsvm example with gaussian kernel | 11:05 |
sonne|work | and try to port it to all $LANG | 11:05 |
bluenotebook | Sonne: Alright. That is what I am going for now. | 11:05 |
sonne|work | then you see the pattern | 11:05 |
bluenotebook | What do you mean by saying: "port it to all $LANG"? | 11:06 |
@HeikoS | lambday: yeah sleeping is important ;) | 11:06 |
sonne|work | bluenotebook: ruby, lua, python, csharp, ... | 11:06 |
lambday | HeikoS: with every change that involves tests from the integration test, we should do thing data PR as well? | 11:07 |
Guest49407 | lambday: sonne|work thanks | 11:07 |
bluenotebook | Sonne|work: Ah. okay. | 11:08 |
-!- kislay [~kislay@93.114.45.13] has joined #shogun | 11:09 | |
@HeikoS | lambday: whenever you break the integration tests you have to re-create the file. Then send a PR against data and commit the data version in shogun. travis will then pull the new data (given the pr is merged) | 11:10 |
@HeikoS | ok, gotta run off to work, see you in a bit | 11:10 |
lambday | HeikoS: alright.. thanks man! taking care of it... ciao :) | 11:11 |
-!- HeikoS [~heiko@2.219.252.217] has quit [Quit: Leaving.] | 11:11 | |
-!- bluenotebook [~bluenoteb@115.249.18.25] has quit [Ping timeout: 265 seconds] | 11:16 | |
-!- aroma [0e8b2998@gateway/web/freenode/ip.14.139.41.152] has quit [Ping timeout: 245 seconds] | 11:20 | |
-!- cameo54321 [6f44611c@gateway/web/freenode/ip.111.68.97.28] has joined #shogun | 11:26 | |
-!- Guest49407 [2eda6d58@gateway/web/freenode/ip.46.218.109.88] has quit [Quit: Page closed] | 11:29 | |
-!- HeikoS [~heiko@pat-191-250.internal.eduroam.ucl.ac.uk] has joined #shogun | 11:35 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 11:35 | |
shogun-notifier- | shogun-data: Parijat Mazumdar :master * beba7df / testsuite/python3-tests/metric_lmnn_modular0.txt: https://github.com/shogun-toolbox/shogun-data/commit/beba7df3c1770032166faa88e62fe304934f9567 | 11:48 |
shogun-notifier- | shogun-data: lmnn_modular data changed #1915 | 11:48 |
shogun-notifier- | shogun-data: Fernando Iglesias :master * 6615cf0 / testsuite/python3-tests/metric_lmnn_modular0.txt: https://github.com/shogun-toolbox/shogun-data/commit/6615cf007634595d459853bf4dc6f1a227d2450c | 11:48 |
shogun-notifier- | shogun-data: Merge pull request #35 from mazumdarparijat/master | 11:48 |
shogun-notifier- | shogun-data: | 11:48 |
shogun-notifier- | shogun-data: lmnn_modular data changed #1915 | 11:48 |
-!- besser82 [quassel@fedora/besser82] has quit [Remote host closed the connection] | 11:56 | |
-!- bluenotebook [~bluenoteb@115.249.18.25] has joined #shogun | 12:10 | |
-!- besser82 [quassel@fedora/besser82] has joined #shogun | 12:20 | |
-!- mode/#shogun [+o besser82] by ChanServ | 12:20 | |
lambday | HeikoS: hi I am not getting it ... inside data/ I see no change :( I rebased the data just now and added the files which could be changed due to my PR but it doesn't get committed | 12:28 |
-!- dhruv13 [~dhruv13@triband-mum-120.61.113.206.mtnl.net.in] has quit [Quit: Leaving] | 12:37 | |
-!- kislay [~kislay@93.114.45.13] has quit [Read error: Connection reset by peer] | 12:42 | |
lambday | ah I have to run generate_testdata.py ain't I :-/ | 12:46 |
-!- dhruv13 [~dhruv13@triband-mum-120.61.113.206.mtnl.net.in] has joined #shogun | 12:47 | |
@besser82 | shiatz! ROFL --> https://www.youtube.com/watch?v=tieA5wfcgH4 | 13:17 |
@besser82 | ^^^ NOT SAFE FOR WORKZZZ!!!! | 13:17 |
lambday | lol | 13:24 |
-!- sunil_ [0e8bd057@gateway/web/freenode/ip.14.139.208.87] has joined #shogun | 13:41 | |
-!- prams [~prams@122.167.127.125] has joined #shogun | 13:41 | |
-!- sunil_ [0e8bd057@gateway/web/freenode/ip.14.139.208.87] has left #shogun [] | 13:45 | |
sonne|work | lambday: did you git add data in shogun git? | 13:49 |
lambday | sonne|work: umm no... | 13:50 |
sonne|work | lambday: well then git doesn't know that you want the newer version of the data submodule | 13:50 |
sonne|work | you have to commit / push that hash too | 13:51 |
-!- vaishaks [~vaishak@1.186.131.85] has quit [Ping timeout: 240 seconds] | 13:52 | |
lambday | sonne|work: okay.. but shouldn't I send a PR against data as well? | 13:52 |
sonne|work | lambday: yeah that is a requirement for the git add data in the shogun repo | 13:53 |
sonne|work | and needs to be merged first | 13:53 |
lambday | sonne|work: that's where I am confused.. in data I don't see any change, so what should I commit? the last change was due to parijat's PR it seems | 13:54 |
sonne|work | ?! | 13:54 |
sonne|work | when you do git submodule update | 13:55 |
sonne|work | lambday: then all good? | 13:55 |
lambday | yep | 13:55 |
-!- prams [~prams@122.167.127.125] has left #shogun [] | 13:55 | |
lambday | sonne|work: I changed the origin url for data/ to my shogun-data fork and kept shogun-toolbox/shogun-data.git as upstream... so that I'd be able to push to my fork.. | 13:58 |
lambday | sonne|work: that won't cause trouble, would it? | 13:58 |
sonne|work | lambday: well don't change origin | 14:04 |
sonne|work | lambday: *add* yours | 14:04 |
sonne|work | I mean just add a new remote lambday or so | 14:04 |
lambday | sonne|work: okay... alright | 14:05 |
lambday | testing | 14:05 |
lambday | sonne|work: added... then did a git submodule update.. seems fine so far | 14:07 |
dhruv13 | HeikoS: hey! I got the fequals function to work with the SerializationJSON* tests | 14:08 |
dhruv13 | HeikoS: https://github.com/shogun-toolbox/shogun/pull/1896 | 14:09 |
-!- kislay [~kislay@93.114.45.13] has joined #shogun | 14:20 | |
sonne|work | dhruv13: so what did you change? | 14:21 |
sonne|work | dhruv13: tolerante - rename it to epsilon | 14:21 |
dhruv13 | added a bool tolerant | 14:21 |
dhruv13 | epsilon is the accuracy parameter | 14:22 |
sonne|work | ahh indeed but wouldn't epsilon = 0 be the same? | 14:22 |
sonne|work | I mean no need for a bool? | 14:22 |
dhruv13 | tolerant switches to epsilon checking, rather than the strict checking suggested here - http://floating-point-gui.de/errors/comparison/ | 14:24 |
dhruv13 | nope, 0 epsilon wouldn't help | 14:24 |
sonne|work | dhruv13: but why don't you use epsilon=1e-7 for json then? | 14:25 |
dhruv13 | 0 epsilon would be the opposite, it would enforce strictest possible checking | 14:25 |
dhruv13 | because, according to the link above, the interpretation for epsilon changes | 14:26 |
dhruv13 | it is now the relative error being checked, rather than difference error | 14:26 |
dhruv13 | ie. diff/(absA + absB) < eps | 14:26 |
dhruv13 | rather than diff < eps | 14:27 |
sonne|work | dhruv13: yeah but still it should work | 14:27 |
dhruv13 | this should make it clear, maybe - http://realtimecollisiondetection.net/blog/?p=89 | 14:28 |
-!- cameo54321 [6f44611c@gateway/web/freenode/ip.111.68.97.28] has quit [Ping timeout: 245 seconds] | 14:28 | |
dhruv13 | umm, when a = 0.74, b = 0.74, diff = 0.000 but diff != 0 implies a difference of the order of FLT_MIN | 14:29 |
dhruv13 | strict checking should not let a == b | 14:30 |
dhruv13 | but the tolerant one will | 14:30 |
dhruv13 | hence the bool | 14:30 |
dhruv13 | FLT_MIN is of the order of 1E-37 | 14:31 |
-!- vaishaks [~vaishak@1.186.131.85] has joined #shogun | 14:37 | |
-!- pirosb3 [8da3ee72@gateway/web/freenode/ip.141.163.238.114] has joined #shogun | 14:42 | |
dhruv13 | sonne|work: eps = 1E-7 works when tolerant = true | 14:44 |
sonne|work | and when tolerant == false? | 14:45 |
pirosb3 | Hi, my name is Daniel. I wanted to ask some questions regarding SoC. Is there anyone available for a chat? Thanks! | 14:46 |
sonne|work | pirosb3: just ask! | 14:47 |
dhruv13 | doesn't, because strict checking requires equality only when diff is in the order of (FLT_MIN * Machine_epsilon)... please check the link | 14:47 |
dhruv13 | ie. when the difference is a denormal floating point number | 14:48 |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 14:48 | |
@besser82 | pirosb3: What's your question? | 14:49 |
-!- kislay [~kislay@93.114.45.13] has quit [Ping timeout: 264 seconds] | 14:50 | |
pirosb3 | OK :) I am currently a 3rd year comp sci student at Plymouth University. Since last year I am very interested in ML, this has made my self-learn Neural Networks (MLP) and Naive Bayes. I have also made a small implementation using CUDA cards | 14:50 |
pirosb3 | I want to use SoC to broaden my skills and contribute to open source. | 14:51 |
@besser82 | pirosb3: sounds interessting :D Which GSoC-Project you are interessted in, especially? | 14:51 |
pirosb3 | I am a fast learner, but I have to be honest: my math background is not the best ;) my question is: If I can prove you that I can learn, could I be a potential candidate? I would love to work on shogun | 14:52 |
-!- prams [~prams@122.167.127.125] has joined #shogun | 14:53 | |
@besser82 | pirosb3: Sure! Why not :D | 14:54 |
sonne|work | pirosb3: yes sure - just start contributing | 14:54 |
pirosb3 | besser82: I was thinking of Shogun, but I wanted to discuss if my skills can be of any help to the project. I know I would be able to learn a lot from Shogun, but I want to make sure I can give back to the community | 14:54 |
sonne|work | and figure out what is of interest to you | 14:54 |
@besser82 | pirosb3: as sonne|work said :D | 14:54 |
@besser82 | pirosb3: come in and find out ;) | 14:55 |
@besser82 | pirosb3: what are your primary skills on IT? | 14:55 |
pirosb3 | I work a lot with concurrent web applications, I am currently building a realtime sentiment analysis system using Twitter. The reason I am saying this is that I am really amazed in how we can visualize data and distribute it using the web as a global platform | 14:56 |
pirosb3 | Thanks for the comments! | 14:56 |
prams | hi, i am having trouble running the tests. I have included gtest header files in usr/local/include . that should be enough to run the tests right.? but I'm getting relocation errors. can you pls help me with this.? | 14:58 |
pirosb3 | I know how to build interactive web applications, and I know that it could be an excellent way to promote the use of Shogun. Given my background, is there something you would see most fit for me? | 14:59 |
dhruv13 | prams: what tests do you want to run? | 14:59 |
dhruv13 | prams: or have you added tests of your own? | 15:00 |
prams | dhruv13 I'm running the test for FastICA and jadeICA algos | 15:01 |
prams | I am working on a fix. i have to write my own tests but I'm not able to run any test.. | 15:01 |
prams | dhruv13 can you help me with this..? | 15:02 |
dhruv13 | in your build folder, open the tests/unit path, and run ./shogun-unit-test --gtest_filter=CFastICA* | 15:03 |
dhruv13 | i seem to be able to run it | 15:03 |
dhruv13 | prams: does that help? | 15:05 |
prams | hmm may be i missed something while setting up my test environment. dhruv13 can you pls go through the procedere once so that i can find out what i missed.. thanks for your help.. :) | 15:06 |
dhruv13 | okay | 15:06 |
dhruv13 | prams: read this - https://github.com/shogun-toolbox/shogun/blob/develop/doc/md/README_cmake.md | 15:07 |
dhruv13 | the testing/debugging options | 15:07 |
-!- kislay [~kislay@93.114.45.13] has joined #shogun | 15:07 | |
@besser82 | pirosb3: you may want to talk to wiking. he's our fancy-pants-stuff guy :D | 15:08 |
prams | dhruv13 ok. let me check | 15:08 |
pirosb3 | besser82: thanks! do you know when I can find him online? would email be better or will I find him on IRC? | 15:09 |
dhruv13 | prams: I think the -DENABLE_TESTING=ON for cmake will do it :) | 15:09 |
@besser82 | pirosb3: we have a "Stammtisch" today, beginning ~ 19:00 UTC. I'm sure you can catch him there | 15:10 |
pirosb3 | besser82: Also, I am a very big docker fan. I know that vigsterkr/shogun image has been created. Wouldn't it be interesting to create a SaaS of Shogun, to be consumed though APIs? | 15:11 |
@besser82 | pirosb3: vigsterkr == wiking ;) | 15:12 |
@besser82 | pirosb3: so that would perfectly fit some of the GSoc tasks mentored by him :D | 15:13 |
@besser82 | pirosb3: like implementing more parallization and possibilities for cloud-intergration | 15:13 |
@besser82 | pirosb3: you have any concrete ideas which platform to use for SaaS? | 15:13 |
pirosb3 | In this way, people can just use Shogun "out of the box" using a simple API, without having to specifically know about the internals. This would increase availability and visibility for Shogun. | 15:14 |
pirosb3 | Oh cool, so wiking is the person I need to speak to :) | 15:14 |
prams | dhruv13 I did enable testing flag during setup. however i'm running all the test again. I'll come back to you if something breaks :D | 15:15 |
pirosb3 | besser82: do you mean in terms of architecture, or vm virtualization system? | 15:15 |
dhruv13 | prams: okay! :-) | 15:15 |
@besser82 | pirosb3: both ;) | 15:16 |
pirosb3 | if we use docker, we create a simple deployable image, we can use "links" and port forwarding that are supported natively by Docker and are super fast. For big datasets I have some interesting ideas to brainstorm.. | 15:16 |
pirosb3 | besser82: I would use Docker for VM, I cannot say about platform now as I need to investigate. Do you have any suggestions? do you have any resources I should start reading about? | 15:17 |
@besser82 | pirosb3: since openstack && openshift want to go for docker-buildin-support I think that may be a possibility | 15:18 |
@besser82 | pirosb3: i'm just recommending openstack || openshift, because I know some of their developers... | 15:19 |
pirosb3 | Yes I think so. There is so much to start reading about, and I am just in the process of installing Shogun now. I will be going through the examples section and will do entry level bugfixing, as stated in the guide. | 15:20 |
@besser82 | pirosb3: which Os you are on, btw? | 15:21 |
pirosb3 | besser82: I will also post on the newsgroup, to get to know the community better | 15:21 |
@besser82 | pirosb3: would be a great idea :D | 15:21 |
pirosb3 | besser82: OSX currently, but I can virtualize though VM. Docker is a bit buggy on OSX. The best solution is OSX -> VM -> Docker. | 15:22 |
pirosb3 | besser82: are you in Berlin? | 15:22 |
-!- dhruv13_ [~dhruv13@triband-mum-120.61.112.43.mtnl.net.in] has joined #shogun | 15:22 | |
@besser82 | pirosb3: unfortunately no, im near Hannover | 15:23 |
pirosb3 | oh okay, was in Berlin yesterday for the first time. Really amazing place. Never been to Hannover but I have friends. | 15:24 |
@besser82 | pirosb3: Berlin is a fantastic city :D Hannover is just a dirty place... | 15:25 |
-!- dhruv13 [~dhruv13@triband-mum-120.61.113.206.mtnl.net.in] has quit [Ping timeout: 244 seconds] | 15:26 | |
pirosb3 | besser82: Ahah! ok | 15:29 |
pirosb3 | besser82: can I participate at the Stammtisch? would be good to speak to him and hear what other people in the community are doing | 15:29 |
@besser82 | pirosb3: yes, sure! Just come along :D | 15:30 |
@besser82 | pirosb3: the stammtisch is a virtual one on irc, btw. ;) | 15:30 |
pirosb3 | Sounds great ;) | 15:31 |
pirosb3 | besser82: Anything else you would reccomend me to look into (other than what was already written in Shogun SoC page) ? | 15:33 |
dhruv13_ | Travis is not building my latest commit/PR | 15:36 |
@besser82 | pirosb3: things on that page should be enough to start, I think... The rest || additonal stuff will be coming piece by piece when "playing" araound with shogun | 15:36 |
dhruv13_ | any way to fix this? | 15:36 |
pirosb3 | Ok great. besser82 sonne|work thanks for your help. see you later! auf wiedersehen | 15:41 |
@besser82 | pirosb3: you're welcome :D Bis später ;) | 15:41 |
-!- vaishaks [~vaishak@1.186.131.85] has quit [Ping timeout: 252 seconds] | 15:42 | |
dhruv13_ | HeikoS: the travis build for my latest commit has not started | 15:47 |
dhruv13_ | HeikoS: how to fix that? | 15:47 |
-!- radgaonkar [~rad@desktop-242.kma.informatik.tu-darmstadt.de] has joined #shogun | 15:55 | |
-!- radgaonkar [~rad@desktop-242.kma.informatik.tu-darmstadt.de] has quit [Quit: leaving] | 15:58 | |
-!- vaishaks [~vaishak@1.186.131.85] has joined #shogun | 16:00 | |
-!- radgaonkar [~rad@desktop-242.kma.informatik.tu-darmstadt.de] has joined #shogun | 16:10 | |
-!- pirosb3 [8da3ee72@gateway/web/freenode/ip.141.163.238.114] has quit [Ping timeout: 245 seconds] | 16:27 | |
-!- kislay [~kislay@93.114.45.13] has quit [Ping timeout: 244 seconds] | 16:32 | |
-!- vaishaks [~vaishak@1.186.131.85] has quit [Ping timeout: 244 seconds] | 16:34 | |
-!- vaishaks [~vaishak@1.186.131.85] has joined #shogun | 16:35 | |
-!- kislay [~kislay@93.114.45.13] has joined #shogun | 16:46 | |
-!- vaishaks [~vaishak@1.186.131.85] has quit [Ping timeout: 264 seconds] | 16:48 | |
-!- kislay [~kislay@93.114.45.13] has quit [Read error: Connection reset by peer] | 16:48 | |
-!- vaishaks [~vaishak@1.186.131.85] has joined #shogun | 16:49 | |
dhruv13_ | sonney2k: travis isn't running a build test on my latest PR: https://github.com/shogun-toolbox/shogun/pull/1896 | 16:53 |
dhruv13_ | sonney2k: why's that? | 16:53 |
-!- kislay [~kislay@93.114.45.13] has joined #shogun | 17:03 | |
-!- cameo54321 [27295846@gateway/web/freenode/ip.39.41.88.70] has joined #shogun | 17:06 | |
-!- batman is now known as makokal | 17:09 | |
-!- vaishaks [~vaishak@1.186.131.85] has quit [Ping timeout: 240 seconds] | 17:11 | |
-!- vaishaks [~vaishak@1.186.131.85] has joined #shogun | 17:17 | |
-!- vaishaks [~vaishak@1.186.131.85] has quit [Ping timeout: 241 seconds] | 17:26 | |
-!- prateekp [73f77fda@gateway/web/freenode/ip.115.247.127.218] has joined #shogun | 17:29 | |
-!- prateekp [73f77fda@gateway/web/freenode/ip.115.247.127.218] has left #shogun [] | 17:29 | |
-!- cameo54321 [27295846@gateway/web/freenode/ip.39.41.88.70] has quit [Ping timeout: 245 seconds] | 17:32 | |
-!- strenpendulous [~bgdavis@teallite.its.unc.edu] has joined #shogun | 17:42 | |
-!- cameo54321 [27295846@gateway/web/freenode/ip.39.41.88.70] has joined #shogun | 17:47 | |
-!- prams [~prams@122.167.127.125] has left #shogun [] | 17:57 | |
-!- kislay [~kislay@93.114.45.13] has quit [Read error: Connection reset by peer] | 18:04 | |
-!- strenpendulous [~bgdavis@teallite.its.unc.edu] has quit [Quit: Leaving.] | 18:26 | |
-!- radgaonkar [~rad@desktop-242.kma.informatik.tu-darmstadt.de] has quit [Quit: leaving] | 18:35 | |
-!- kislay [~kislay@93.114.45.13] has joined #shogun | 18:52 | |
@sonney2k | dhruv13_, travis tries each and every PR | 19:03 |
-!- zxtx [~zv@c-98-223-196-32.hsd1.in.comcast.net] has quit [Ping timeout: 240 seconds] | 19:06 | |
-!- besser82 [quassel@fedora/besser82] has quit [Quit: freedom, friends, features, first ---> fedoraproject.org] | 19:06 | |
-!- pickle27 [c0008876@gateway/web/freenode/ip.192.0.136.118] has joined #shogun | 19:09 | |
-!- sunil1337 [0e8bd057@gateway/web/freenode/ip.14.139.208.87] has joined #shogun | 19:15 | |
-!- besser82 [quassel@fedora/besser82] has joined #shogun | 19:22 | |
-!- mode/#shogun [+o besser82] by ChanServ | 19:22 | |
-!- besser82 [quassel@fedora/besser82] has quit [Quit: freedom, friends, features, first ---> fedoraproject.org] | 19:26 | |
-!- besser82 [quassel@fedora/besser82] has joined #shogun | 19:30 | |
-!- mode/#shogun [+o besser82] by ChanServ | 19:30 | |
@HeikoS | dhruv13_: travis starts once there is computing timer availabvle :) | 19:30 |
@HeikoS | but I can restart, can you send me a link? | 19:30 |
@HeikoS | lambday: around? | 19:31 |
lambday | HeikoS: yeah | 19:36 |
lambday | HeikoS: just came back! | 19:36 |
-!- thoralf [~thoralf@91-65-138-84-dynip.superkabel.de] has joined #shogun | 19:36 | |
@HeikoS | lambday: ah good | 19:36 |
thoralf | Heyho. | 19:36 |
@HeikoS | I just talked to dino for a while | 19:36 |
@HeikoS | thoralf: hi! :) | 19:36 |
lambday | HeikoS: the PR seems green | 19:36 |
@HeikoS | lambday: and there should not be another class in between | 19:36 |
@HeikoS | lambday: in fact, the way two-sample test and independence test store data is different | 19:36 |
lambday | HeikoS: umm.. why so? how about three distribution one? | 19:36 |
@HeikoS | lambday: two sample can store both P and Q in the same data object, but independence test has to store them in different objects (they might be from different domains=) | 19:37 |
@HeikoS | lambday: so you can leave it as it was | 19:37 |
@HeikoS | CTestStatistic->CTwoSampleTest-> .... (one feature object m_p_and_q with convenience constructors) | 19:38 |
@HeikoS | CTestStatistic->CIndependenceTest-> .... (two feature objects) | 19:38 |
lambday | HeikoS: umm... so, there would be no p_and_q in independence one? | 19:38 |
@HeikoS | lambday: and then the permutation test for independence just permutes one feature objects rather than the index subset only on one | 19:38 |
@HeikoS | lambday: exactly | 19:38 |
lambday | HeikoS: actually I was thinking what is the need of having two different kernels when we init them with the same set of features | 19:39 |
lambday | for independence test | 19:39 |
@HeikoS | lambday: yeah, there need to be two kernels for CKernelIndependenceTest | 19:39 |
@HeikoS | you are exactly right | 19:39 |
@HeikoS | lambday: the independence tests that I did were quick hacks, thats why there all those mistakes ;) | 19:39 |
@HeikoS | lambday: I did not know they stuff too well back then | 19:39 |
lambday | HeikoS: but those kernels shouldn't compute the kernel function *between* the different samples | 19:39 |
@HeikoS | lambday: no for independence test the kernel is only on P and Q seperately | 19:40 |
lambday | HeikoS: and still I know nothing :( | 19:40 |
@HeikoS | and then the kernel matrices are multiplied | 19:40 |
@HeikoS | lambday: dont worry :) | 19:40 |
@HeikoS | lambday: dino noticed all that, he will be a great help | 19:40 |
@HeikoS | lambday: but that doesnt concern the current PR right? | 19:40 |
lambday | HeikoS: ah this is getting really interesting | 19:40 |
@HeikoS | lambday: because I will merge that now | 19:40 |
lambday | HeikoS: nah as I promised, the current one is just for renaming | 19:40 |
@HeikoS | lambday: great then | 19:41 |
lambday | and some minor documentation check | 19:41 |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 19:41 | |
shogun-notifier- | shogun: lambday :develop * f70fce0 / / (31 files): https://github.com/shogun-toolbox/shogun/commit/f70fce02466d9fd86401678d94430838076e1c54 | 19:41 |
shogun-notifier- | shogun: rename bootstrap in test statistic framework | 19:41 |
shogun-notifier- | shogun: Heiko Strathmann :develop * db13b04 / / (31 files): https://github.com/shogun-toolbox/shogun/commit/db13b044546804b8e29b64958c74b284f9eb9616 | 19:41 |
shogun-notifier- | shogun: Merge pull request #1921 from lambday/develop | 19:41 |
shogun-notifier- | shogun: | 19:41 |
shogun-notifier- | shogun: rename bootstrap in test statistic framework | 19:41 |
lambday | HeikoS: I really really should read the papers within a day or two | 19:41 |
lambday | at least the mmd and hsic one | 19:41 |
@HeikoS | lambday: read my MSc dissertation | 19:41 |
@HeikoS | for mmd | 19:41 |
@HeikoS | has a good summary | 19:42 |
@HeikoS | and the "a kernel two sample test" JMLR paper | 19:42 |
lambday | HeikoS: could you please give me the link | 19:42 |
lambday | of your dissertation | 19:42 |
lambday | alrighty | 19:42 |
@HeikoS | lambday: http://herrstrathmann.de/research/86-master-thesis.html | 19:43 |
lambday | HeikoS: thanks a lot! :) | 19:43 |
-!- dino_ [9052bdeb@gateway/web/freenode/ip.144.82.189.235] has joined #shogun | 19:43 | |
-!- zxtx [~zv@129-79-241-148.dhcp-bl.indiana.edu] has joined #shogun | 19:44 | |
lambday | HeikoS: just a min... be right back after a quick call | 19:46 |
-!- pirosb3 [8da3303c@gateway/web/freenode/ip.141.163.48.60] has joined #shogun | 19:52 | |
@besser82 | pirosb3: wb! :D | 19:53 |
pirosb3 | Hey besser82! | 19:54 |
@besser82 | pirosb3: and here, take some virtual beer :-P | 19:54 |
lambday | HeikoS: back! | 19:54 |
pirosb3 | Oh I would love some, thanks :) next beer is on me! | 19:54 |
pirosb3 | besser82: I am compiling Shogun. to develop on it, I should just be doing "make" after cmake right? (not doing "make install") | 19:55 |
lambday | HeikoS: alright... so I will change the merged features thing for independence test in the next PR.... but you sure that having no class in between (which means that we'll either have to push the compute_p_value and compute_threshold to the common parent CTestStatistic or we keep duplicated code) would fit in future? | 19:55 |
@HeikoS | no duplicated | 19:57 |
@HeikoS | lambday: it *is* different after all for two-sample (merge + permute) and indepdendence (permute P only) | 19:57 |
lambday | HeikoS: but the common code we're talking about isn't that permutation part, right? | 19:58 |
lambday | HeikoS: let me check.. just a min | 19:58 |
@besser82 | pirosb3: if you are developing on it, I'd recommend to clean build-dir and re-run cmake && make after every change / git fetch | 19:58 |
pirosb3 | perfect. Thanks besser82 | 19:59 |
@besser82 | pirosb3: np :D | 19:59 |
@HeikoS | lambday: ? | 20:00 |
-!- dino_ [9052bdeb@gateway/web/freenode/ip.144.82.189.235] has quit [Quit: Page closed] | 20:01 | |
-!- HeikoS [~heiko@pat-191-250.internal.eduroam.ucl.ac.uk] has left #shogun [] | 20:01 | |
lambday | HeikoS: I am talking about compute_p_value and compute_threshold.. which, of course, relies on sample_null (which is implementation specific) but that would still be virtual in parent class.. so as soon as subclasses provide the definitions of sample_null (based on permuting both for two-sample or permuting just one for independence test) its good to go | 20:01 |
lambday | errr! | 20:01 |
-!- HeikoS [~heiko@pat-191-250.internal.eduroam.ucl.ac.uk] has joined #shogun | 20:01 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 20:01 | |
lambday | HeikoS: wb | 20:01 |
@HeikoS | wie kommts? | 20:01 |
lambday | HeikoS: I am talking about compute_p_value and compute_threshold.. which, of course, relies on sample_null (which is implementation specific) but that would still be virtual in parent class.. so as soon as subclasses provide the definitions of sample_null (based on permuting both for two-sample or permuting just one for independence test) its good to go | 20:02 |
@HeikoS | yeah the interface is defined in CTestStatistic | 20:02 |
@HeikoS | but the implementation is in abstract class CTwoSampleTest CIndependenceTEst | 20:02 |
@HeikoS | and the kernel classes re-implement that again with kernel caching if desirded | 20:02 |
lambday | HeikoS: for sample_null, you mean, right? | 20:02 |
@HeikoS | yeah | 20:02 |
lambday | HeikoS: I am talking about compute_p_value and compute_threshold - these two methods | 20:03 |
@HeikoS | lambday: ah sorry, not that is common | 20:03 |
@HeikoS | I guess that can be in CTestStatistic right? | 20:03 |
lambday | HeikoS: yeah so we can pull these two inside test statistic | 20:03 |
@HeikoS | because the code only relies on the abstract interface of CTestStatistic | 20:03 |
lambday | HeikoS: I think so | 20:03 |
@HeikoS | lambday: am I right?, not sure currently | 20:03 |
lambday | HeikoS: yeah - unless there are implementations of CTestStatistic for which these implementations would absolutely make no sense! | 20:04 |
lambday | which I am not sure atm | 20:04 |
@HeikoS | lambday: nono, every hypothesis test has those :) | 20:04 |
@besser82 | wiking: ping? | 20:05 |
@HeikoS | lambday: maybe we can even think of ranaming that class to CHypothesisTest | 20:05 |
lambday | HeikoS: alright then! it all depends on the implementation of different sample_null | 20:05 |
lambday | HeikoS: test statistic one? | 20:05 |
@besser82 | wiking: pirosb3 wants to get in touch with you... | 20:05 |
@HeikoS | and CTwoSampleTest CIndependenceTest, CKernelTwoSampleTest, CKernelIndependenceTest | 20:05 |
@HeikoS | lambday: yeah | 20:05 |
lambday | HeikoS: umm yeah that sounds good! | 20:06 |
lambday | HeikoS: so it would have no problem with future implementations you think? | 20:06 |
lambday | pulling the p-value and threshold computation upward to base class? | 20:06 |
pirosb3 | hi wiking, I wanted to speak to you about an interesting idea I have for SoC. It would start from the work you did with Docker. Let me know when you have a sec :) | 20:06 |
@besser82 | Hey Folks! Welcome to the Stammtisch :D | 20:06 |
@HeikoS | lambday: let me check the code, 1 min | 20:07 |
@besser82 | thoralf: hey! 'sup? | 20:07 |
pirosb3 | Hi guys :) | 20:07 |
lambday | besser82: hello, sir :) | 20:07 |
thoralf | besser82: Man! SHOGUN 's up! :) | 20:07 |
@besser82 | lambday: hey! How is it going? | 20:07 |
@HeikoS | besser82: yeah Stammtisch :) | 20:08 |
@besser82 | HeikoS: yeah! Time to pimp my task-descriptions :D | 20:08 |
thoralf | besser82: Uh. | 20:08 |
@HeikoS | wiking, sonney2k, sonne|work whats up? Important things to do? :) | 20:08 |
thoralf | Reminds me I should do the same. | 20:08 |
@HeikoS | besser82, thoralf yeah gogog :) | 20:08 |
@wiking | besser82: pirosb3 ? | 20:08 |
lambday | besser82: going good man! you'll be mentoring the job framework one, right? | 20:08 |
@wiking | HeikoS: pong | 20:08 |
@wiking | sonney2k: i'll get into mex fixing | 20:09 |
@HeikoS | wiking: ping :) | 20:09 |
@HeikoS | wiking: any ideas on this notebook tool? | 20:09 |
pirosb3 | Hi wiking. Pleased to meet you | 20:09 |
cameo54321 | Hi guys! Arsalan here. | 20:09 |
cameo54321 | I am new to shogun | 20:09 |
pirosb3 | hi cameo54321 | 20:09 |
@wiking | HeikoS: what notebook tool? :) sorry i've just joined | 20:09 |
@wiking | pirosb3: hey... | 20:10 |
@wiking | pirosb3: how can i help u? | 20:10 |
cameo54321 | And shoguns seesm to have quite an intersting community :-) | 20:10 |
@HeikoS | wiking: rendering notebooks from travis | 20:10 |
cameo54321 | shogun* | 20:10 |
thoralf | cameo54321: It is. ;) | 20:10 |
@wiking | HeikoS: ah no... i'll add the new bot... | 20:10 |
@HeikoS | cameo54321: indeed! | 20:10 |
@HeikoS | wiking: great! | 20:10 |
pirosb3 | I pulled down your docker image | 20:10 |
@wiking | HeikoS: it's going to be buildbot based | 20:10 |
cameo54321 | :-) | 20:10 |
@HeikoS | lambday: I think both methods can go up | 20:10 |
pirosb3 | it works perfectly, I was thinking | 20:10 |
@wiking | pirosb3: ok so that needs a lot of work... actually it's better to base your work on the Dockerfile in the repo | 20:10 |
@HeikoS | only refers to bootstrap_null | 20:11 |
lambday | HeikoS: alrighty | 20:11 |
@wiking | pirosb3: so you wanna work on cloud extensions? | 20:11 |
cameo54321 | I posted on the mailing list today on MCMC thing | 20:11 |
lambday | HeikoS: then my next PR is already ready I guess - just checking a bit | 20:11 |
@HeikoS | cameo54321: hi, I am about to answer | 20:11 |
pirosb3 | what if we could expand it, and make Shogun a Saas | 20:11 |
lambday | HeikoS: oh no! I need to use two different features for independence test | 20:11 |
@HeikoS | cameo54321: but have a look at the python code and the class diagram we wrote | 20:11 |
lambday | HeikoS: checking... shouldn't take long | 20:12 |
@HeikoS | cameo54321: also, have a look at stan and try to understand some internals | 20:12 |
@wiking | pirosb3: yeah well that was the idea with the cloud shogun | 20:12 |
@HeikoS | cameo54321: but only basics, more important is the class diagram | 20:12 |
cameo54321 | HeikoS: That would be great | 20:12 |
@wiking | pirosb3: of course one thing of it would be the whole notebooking stuff | 20:12 |
cameo54321 | Yes I have seen the class diagram | 20:12 |
@wiking | pirosb3: of course if u want it as a saas, then you'll have to start defining a standard api | 20:12 |
@HeikoS | lambday: yeah two feature objects, thats a a bigger change as it will affect bootstrap null | 20:12 |
pirosb3 | wiking: exactly. we can make any application communicate with Shogun | 20:13 |
pirosb3 | though Docker links | 20:13 |
dhruv13_ | hey! sorry, was afk... | 20:13 |
@wiking | pirosb3: u mean via REST api right? | 20:13 |
pirosb3 | it would be really cool to define a high level interface that anyone could interact with, without knowing about the internals | 20:14 |
pirosb3 | wiking: sure, but then how would you handle big data | 20:14 |
@wiking | pirosb3: mmm then i dont really get how do u want to do SaaS? | 20:14 |
pirosb3 | would REST still be sufficient? | 20:14 |
@wiking | could u elaborate | 20:14 |
pirosb3 | wiking: I am thinking of an idea.. | 20:14 |
pirosb3 | So you can link folders in Docker | 20:15 |
pirosb3 | it would be cool to have a way Shogun can read files from a folder | 20:15 |
@wiking | pirosb3: that's already done | 20:15 |
pirosb3 | and process them. We would still need a restful API for interacting with the service | 20:16 |
pirosb3 | oh cool! is there anywhere I can find the specs? | 20:16 |
@wiking | pirosb3: well you should start checking the io api | 20:16 |
pirosb3 | ok perfect, I am in the process of building shogun as we speak | 20:17 |
@wiking | pirosb3: LibSVMFile.h can read libsvm formated file... | 20:17 |
pirosb3 | oh no! my build just failed | 20:17 |
@wiking | CSVFile.h can read csv file | 20:17 |
@wiking | and so on | 20:17 |
pirosb3 | but how does it work with Docker? | 20:18 |
@wiking | pirosb3: well in docker you have a full shogun | 20:18 |
@wiking | so it can do whatever u want within docker | 20:18 |
pirosb3 | do you just drop the files there and then the process is automated? | 20:18 |
@sonney2k | wiking, mex mex hurray! | 20:18 |
@besser82 | sonney2k: hey! 'sup? | 20:19 |
shogun-notifier- | shogun-web: Kevin Hughes :master * 599f7c1 / pages/views.py,static/css/styles.css: https://github.com/shogun-toolbox/shogun-web/commit/599f7c120e054960c863a8f48c9eb5e33c98adc9 | 20:19 |
shogun-notifier- | shogun-web: fix zindex of carousel indicators and fix page view error | 20:19 |
shogun-notifier- | shogun-web: Kevin Hughes :master * 9f515fc / static/css/styles.css: https://github.com/shogun-toolbox/shogun-web/commit/9f515fc5a3b51265a24c8e0c60cad79437b8d0f6 | 20:19 |
shogun-notifier- | shogun-web: Merge pull request #55 from pickle27/master | 20:19 |
shogun-notifier- | shogun-web: | 20:19 |
shogun-notifier- | shogun-web: fix zindex of carousel indicators and fix page view error | 20:19 |
@besser82 | pirosb3: no, currently not :( You need to do lauch things manually afaik... | 20:19 |
@sonney2k | besser82, I uploaded the new .deb based on your congfu script :) | 20:20 |
@wiking | pirosb3: i mean i dont know what would u like to do when a file is being uploaded to a docker image | 20:20 |
@wiking | pirosb3: shogun is huge with a lot of algos | 20:20 |
@wiking | pirosb3: so it really depends on the use-case | 20:20 |
pirosb3 | wiking: besser82 what about something like this | 20:20 |
pirosb3 | https://developers.google.com/prediction/docs/hello_world | 20:21 |
dhruv13_ | HeikoS: Really stupid typo in my latest commit | 20:21 |
dhruv13_ | sorry | 20:21 |
@besser82 | sonney2k: Still fighting with testsuite on ARMv7hf for F20 and SWIGZZ && Octave on rawhide | 20:21 |
@HeikoS | dhruv13_: dont worry, I will very soon! | 20:21 |
@wiking | pirosb3: this is the REST api i was talking about | 20:21 |
@sonney2k | besser82, swig2.12 fixes the octave3.8 issue | 20:21 |
@wiking | pirosb3: if u want that that's a different story | 20:21 |
dhruv13_ | HeikoS: the travis build started :) | 20:21 |
-!- votjakovr [~votjakovr@188.134.46.30] has joined #shogun | 20:21 | |
dhruv13_ | HeikoS: but i made a few cosmetic changes, and made a type | 20:21 |
@besser82 | sonney2k: I know, but we still need to update to that ver... | 20:21 |
@wiking | pirosb3: for that u have to start define REST api and link it with the current algos | 20:22 |
dhruv13_ | HeikoS: * typo | 20:22 |
dhruv13_ | HeikoS: lol | 20:22 |
@wiking | pirosb3: but of course the best would be that the mapping is somehow autogen | 20:22 |
@besser82 | sonney2k: testsuite of swig chokes for some unknown reason... | 20:22 |
pirosb3 | wiking: exactly, I was thinking something like that | 20:22 |
@wiking | pirosb3: e.g. if a new algorithm is being added then it's automatically revealed via the rest api | 20:22 |
@wiking | pirosb3: this can be done quite easily | 20:22 |
@wiking | of course needs time | 20:23 |
pirosb3 | wiking: yes very cool! | 20:23 |
shogun-notifier- | shogun-web: Kevin Hughes :master * 8eca093 / pages/admin.py: https://github.com/shogun-toolbox/shogun-web/commit/8eca09370fefe2dfd97761bc2c6a8b62d4cc5bfa | 20:23 |
shogun-notifier- | shogun-web: remove page and subpage from admin | 20:23 |
shogun-notifier- | shogun-web: Kevin Hughes :master * e0aa7b6 / pages/admin.py: https://github.com/shogun-toolbox/shogun-web/commit/e0aa7b6b885870b1818f97ac47b7bc243ba34c70 | 20:23 |
shogun-notifier- | shogun-web: Merge pull request #56 from pickle27/master | 20:23 |
lambday | HeikoS: should I rename TestStatistic with HypothesisTest? | 20:23 |
shogun-notifier- | shogun-web: | 20:23 |
shogun-notifier- | shogun-web: remove page and subpage from admin | 20:23 |
pirosb3 | wiking: would you suggest me a place to start? | 20:23 |
@wiking | pirosb3: but, as shogun has a good hierarchy of classes, it's not that hard to actually create a high-level REST api | 20:23 |
pirosb3 | after having done all the instructions in Soc page | 20:23 |
pirosb3 | that's really good then. I really thing this is something that would be really good to have | 20:24 |
@wiking | pirosb3: well the thing is that we would actually need to work on planning how to do this. e.g. what serves the rest api.... can we do this like a http server independent engine .... etc. | 20:24 |
@HeikoS | lambday: yeah I guess thats best | 20:25 |
@wiking | pirosb3: or are we just use Flask as a webserver and then somehow generate the high-level api | 20:25 |
@wiking | pirosb3: so that's quite some things to spec out | 20:25 |
@HeikoS | lambday: use some refactoring magic | 20:25 |
pirosb3 | Docker is really changing how services interact with each other, there are many indipendent services for databases, frameworks... and here is where Shogun comes in | 20:25 |
@HeikoS | lambday: otherwise too much work :) | 20:25 |
@sonney2k | pirosb3, do you want the latest changes deployed? | 20:25 |
@wiking | pirosb3: but before that you should start working with shogun itself | 20:25 |
@sonney2k | err | 20:25 |
lambday | HeikoS: yeah, grep+find I guess :D | 20:26 |
@sonney2k | pickle27, do you want the latest changes deployed? | 20:26 |
thoralf | wiking: Hmm. Adding more and more technologies might reduce the time we have left for algorithms... | 20:26 |
pirosb3 | we can provide a machine learning service | 20:26 |
@wiking | pirosb3: and work on some of the enterance tasks... | 20:26 |
pirosb3 | sonney2k: yes please | 20:26 |
pickle27 | sonney2k: I have a few more that will come soon | 20:26 |
@wiking | pirosb3: for example in this case we would like to have an automagically build docker image + repository (until the problem with trusted builds is not being resolves) | 20:26 |
pirosb3 | wiking: of course, I will be doing everything that is said in the SoC docs | 20:26 |
pickle27 | sonney2k: also yay the exceptions stopped! I think we just had a bunch of dead links that got cached to some web crawlers somehow | 20:26 |
@wiking | pirosb3: that's like issue 0) at this stage | 20:27 |
@sonney2k | pickle27, well you return a 404 right? | 20:27 |
pickle27 | that too | 20:27 |
@sonney2k | pickle27, if you see any chance to speed up the page now would be the time :) | 20:27 |
pickle27 | but I actually think I somehow didn't merge that yet | 20:27 |
@wiking | pirosb3: i'll think about more entrance task like problems in this area and i'll put them up to the issues | 20:27 |
pirosb3 | wiking: thats great | 20:28 |
pirosb3 | wiking: is your Docker image built from a Dockerfile? | 20:28 |
@wiking | pirosb3: but first i'll have to think about how we could do this high-level rest API... as that'll be some time to figure out... | 20:28 |
@wiking | pirosb3: not exactly... that Dockerfile is meant for trusted-builds of docker (index.docker.io) | 20:28 |
pirosb3 | wiking: I would be really happy to brainstorm on it together with everyone else | 20:29 |
@wiking | pirosb3: i think mailinglist would be the best place to do this | 20:29 |
sunil1337 | hi guys sunil here | 20:29 |
pirosb3 | perfect wiking. I will start by writing on the ML. Let's see how the community reacts. | 20:30 |
pirosb3 | but first, I really need to get some experience with Shogun! | 20:30 |
@sonney2k | HeikoS, we have a problem with the fequals... as dhruv13_ pointed out NaN != NaN ... which is not what we want when we want to check if two objects are equal by deep comparison | 20:30 |
@wiking | pirosb3: yeah just drop some lines there and let's see how that rumbles | 20:30 |
@wiking | pirosb3: yeah a good way to start is to try to build shogun on your own ... | 20:31 |
@HeikoS | sonney2k, dhruv13_ I see, I am about to check this., But I would say just check NaN by hand then, in fequals, it should give false | 20:31 |
pirosb3 | Also wiking. I am very passionate about machine learning, but I have to be honest: I do not have too much experience :) | 20:31 |
-!- tomy18 [~alex@p6.eregie.pub.ro] has joined #shogun | 20:31 | |
-!- radgaonkar [~rad@p4FC73B2E.dip0.t-ipconnect.de] has joined #shogun | 20:32 | |
pirosb3 | I have self-taught most of the stuff I know, so I will be very happy to learn by experience as I have always done. Do you think this task will teach me how to apply machine learning models? | 20:32 |
pirosb3 | I really want to increase my knowledge if ML :) | 20:33 |
@sonney2k | HeikoS, no - it should give true for sgobjectA.equals(sgobjectB) | 20:33 |
@HeikoS | sonney2k: NaN should not be equal to anything in a floating point equals method | 20:34 |
@HeikoS | sonney2k: but it should be equal for CSGObject | 20:34 |
@HeikoS | so I would rather check by hand | 20:34 |
-!- kislay [~kislay@93.114.45.13] has quit [Read error: Connection reset by peer] | 20:34 | |
-!- bluenotebook [~bluenoteb@115.249.18.25] has quit [Ping timeout: 264 seconds] | 20:35 | |
pirosb3 | Build is failing at: shogun/base/class_list.cpp:723:101: error: unknown type name 'CPCA' | 20:35 |
@sonney2k | HeikoS, what does that mean - check by hand? | 20:35 |
pirosb3 | Does anyone know what that could be? any hint? | 20:35 |
@HeikoS | sonney2k: should be done if (fequals(a,b) or (a==Nan && b==Nan)) then | 20:36 |
@HeikoS | pirosb3: include guard for lapack/eigen | 20:36 |
@HeikoS | pirosb3: update to latest git, and if the error persists, pls report it (giving all logs) on github | 20:37 |
@besser82 | HeikoS: shouldn't that be in latest checkout already? | 20:37 |
pirosb3 | Great | 20:37 |
@HeikoS | pirosb3: then, install lapack and eigen3, the error should be gone, but we still dont want it, so it would be useful if you reported it | 20:37 |
@HeikoS | besser82: nope | 20:37 |
pirosb3 | I just pulled from develop, no changes | 20:38 |
pirosb3 | I need to install lapack and eigen3 | 20:38 |
@besser82 | HeikoS: kk, i was just asking, because i thought there had been a commit about some eigen/lapack guards from sonney2k | 20:39 |
@HeikoS | besser82: yeah I know, but this is an error when somebody doesnt have lapack installed and shogun should compile without lapack | 20:39 |
dhruv13_ | sonney2k, HeikoS : NAN is essentially unordered, but there is std::isnan to check | 20:39 |
@HeikoS | dhruv13_: whats this tolerant parameter? | 20:40 |
@besser82 | HeikoS: allrighty, thanks for clarification :D | 20:41 |
dhruv13_ | HeikoS: i had to introduce it to handle JSON truncation of really small values | 20:41 |
-!- cameo54321 [27295846@gateway/web/freenode/ip.39.41.88.70] has quit [Ping timeout: 245 seconds] | 20:41 | |
dhruv13_ | HeikoS: which cannot be taken care of by simply introducing a really small epsilon | 20:41 |
@besser82 | pirosb3: I'll try to punch off that lapack/eigen-related error during tomorrow | 20:41 |
@HeikoS | dhruv13_: so just to be clear, NaN should not be the same (as they are unordered, you are right) for fequals. But two CSGObject instances should be equal if they both contain NaN | 20:41 |
@HeikoS | dhruv13_: I see, but cant that be solved via an epsilon 1e-4 or so? | 20:42 |
dhruv13_ | HeikoS: I can handle that as a special case | 20:42 |
@HeikoS | thats what I found | 20:42 |
dhruv13_ | HeikoS: since the checking method uses relative error, the interpretation of epsilon changes | 20:42 |
pirosb3 | no worries besser82 . But isn't Lapack supposed to be removed? https://github.com/shogun-toolbox/shogun/issues/1553 | 20:42 |
@HeikoS | dhruv13_: I see | 20:43 |
@besser82 | pirosb3: maybe somewhen in future... | 20:43 |
dhruv13_ | HeikoS: epsilon as difference error is different from epsilon as relative error | 20:43 |
@besser82 | pirosb3: there was no specific announce, yet... just a proposal | 20:43 |
pirosb3 | oh ok | 20:43 |
@HeikoS | dhruv13_: ah I dont know about this new parameter | 20:44 |
dhruv13_ | HeikoS: the assertions in the unit tests were in the SerializationJSON.* tests | 20:45 |
@HeikoS | it is so dirty to add this just because of JSON | 20:45 |
pirosb3 | Guys, what do you think about making Shogun a service? (with Docker) do you think it could give value? | 20:45 |
@HeikoS | dhruv13_: yeah I know you did not add them, just wanted to make sure anyways | 20:45 |
dhruv13_ | :) | 20:45 |
dhruv13_ | HeikoS: I think it is logical to introduce the tolerant parameter to have linient checking | 20:45 |
pirosb3 | I just want to make sure it's something that make Shogun better | 20:45 |
@HeikoS | dhruv13_: so if you add the tolerant one, pls make sure that it has a default parameter so that the calls stay the same and only change if one does things from JSON | 20:46 |
@sonney2k | HeikoS, well it is not - serialization is a lossy process | 20:46 |
-!- Jey [~chatzilla@2601:4:700:363:8048:a36b:b0fb:5ec8] has joined #shogun | 20:46 | |
dhruv13_ | HeikoS: yeah, the default parameter is false | 20:46 |
@HeikoS | sonney2k: ? | 20:46 |
dhruv13_ | HeikoS: and i suspect we'll need it for other fringe cases | 20:47 |
@sonney2k | HeikoS, well you we work with floats some crude approximations to real numbers | 20:47 |
@HeikoS | sonney2k: I know that, but I dont get what you want to tell me ;) | 20:47 |
@sonney2k | HeikoS, if we want them to be transferred w/o loss we need to export them as strings | 20:47 |
@HeikoS | dhruv13_: ok, fine then! | 20:48 |
-!- kislay [~kislay@93.114.45.13] has joined #shogun | 20:48 | |
dhruv13_ | pirosb3: shogun as a saas sounds like a good idea to me! | 20:49 |
-!- radgaonkar [~rad@p4FC73B2E.dip0.t-ipconnect.de] has quit [Ping timeout: 240 seconds] | 20:49 | |
-!- radgaonkar [~rad@p4FC73B2E.dip0.t-ipconnect.de] has joined #shogun | 20:51 | |
dhruv13_ | HeikoS: what is the difference between expect and assert? | 20:51 |
dhruv13_ | HeikoS: i mean, why prefer expect? | 20:52 |
@sonney2k | HeikoS, I wanted to tell you that we loose some eps precision when serializing floats | 20:52 |
shogun-notifier- | shogun-web: Kevin Hughes :master * 14eb44b / pages/ (4 files): https://github.com/shogun-toolbox/shogun-web/commit/14eb44b62c961da9112d6b4beb61b76076ea8a89 | 20:53 |
shogun-notifier- | shogun-web: remove foreign_key from article | 20:53 |
shogun-notifier- | shogun-web: Kevin Hughes :master * 5f6450b / pages/ (4 files): https://github.com/shogun-toolbox/shogun-web/commit/5f6450bfb0f680746c4b26a5e55760a55d4a570f | 20:53 |
shogun-notifier- | shogun-web: Merge pull request #57 from pickle27/remove_legacy_models | 20:53 |
shogun-notifier- | shogun-web: | 20:53 |
shogun-notifier- | shogun-web: remove foreign_key from article | 20:53 |
pickle27 | sonney2k: can I get a deploy now? | 20:53 |
pickle27 | also can you run the evolution | 20:53 |
pickle27 | and maybe backup the db before hand | 20:54 |
@sonney2k | wiking, any idea why the static and modular interface no longer build on the buildbot? but just libshogun?! | 20:54 |
@sonney2k | pickle27, errm? | 20:54 |
pirosb3 | thanks dhruv13_ . Do you think I will be able to learn more about using different ML models with this task too? | 20:54 |
@sonney2k | pickle27, then rather not today I am too tired already for that | 20:54 |
pickle27 | sonney2k: okay thats cool | 20:54 |
pickle27 | sonney2k: this is mainly some cleanup removing stuff we don't need any more | 20:54 |
pickle27 | so no rush | 20:54 |
pickle27 | but now we be a good time to back up the data | 20:55 |
@sonney2k | pickle27, we have daily backups anyways | 20:56 |
@HeikoS | sonney2k: ok, yeah sure | 20:56 |
pickle27 | okay cool | 20:56 |
@HeikoS | dhruv13_: both assert and expect make the test fail, but assert makes it STOP, where as expect runs to the end and then fails | 20:56 |
dhruv13_ | pirosb3: I think designing the api in an intuitive way will be a challenge.... you will probably have to think a lot about use cases... I'm the same as you though, an aspiring shogun dev ;-) | 20:57 |
@sonney2k | pirosb3, and yes MEGA +1 for shogun as a service :) | 20:57 |
@HeikoS | dhruv13_: but its important to run tests to the end (expect for example ASSERT(pointer-not-null) when you call a method on the pointer later | 20:57 |
pirosb3 | ahahaha ok cool! thanks for the kudos guys :) makes me happy | 20:57 |
@sonney2k | pirosb3, I thought of that already last year but it is work(tm) | 20:57 |
pirosb3 | hey dhruv13_ are you doing SoC too? | 20:58 |
@sonney2k | pirosb3, my approach to this would be to make it solve a couple of standard problems - mostly automagic | 20:58 |
pirosb3 | sonney2k: thats why I want everyone to get involved, there is a lot of brainstorming to do. But I feel it can be done | 20:58 |
dhruv13_ | pirosb3: yeah! last year in college... really keen on shogun this year... | 20:59 |
-!- kislay [~kislay@93.114.45.13] has quit [Ping timeout: 265 seconds] | 20:59 | |
@sonney2k | pirosb3, as in you upload some data , tell it that this is text data or real valued stuff etc etc and then it tries the fast linear / online methods on this | 20:59 |
pirosb3 | nice dhruv13_ :) | 20:59 |
dhruv13_ | :) | 20:59 |
pirosb3 | what if the folder_name could be some sort of identifier? | 20:59 |
@sonney2k | pirosb3, and unfortunately presentation matters - basically every startup (and $BIG_COMPANY) is doing that now | 21:00 |
pirosb3 | twitter_sentiment_analysis/train/xxxxx.svmlight | 21:00 |
@HeikoS | guys, I will be back in a few minutes, getting some food | 21:00 |
pirosb3 | twitter_sentiment_analysis/test/xxxxx.svmlight | 21:00 |
pirosb3 | they you could ping some sort of REST api. GET http://service:8888/trained_models/twitter_sentiment_analysis/status. Which would return something like {status: "training"} | 21:01 |
pirosb3 | once {status: "ready"}, you could do POST http://service:8888/trained_models/twitter_sentiment_analysis/classify with some features | 21:02 |
-!- Jey [~chatzilla@2601:4:700:363:8048:a36b:b0fb:5ec8] has quit [Quit: ChatZilla 0.9.90.1 [Firefox 27.0.1/20140212131424]] | 21:03 | |
pirosb3 | in this way, developer knows just the necessary about the ML service, and no dependency/build necessary for him | 21:04 |
pirosb3 | 100% profit ;) | 21:04 |
dhruv13_ | pirosb3: how would you point the ML service to data efficiently? | 21:05 |
@besser82 | pirosb3: sounds awesome :D | 21:05 |
pirosb3 | dhruv13_: so I was thinking there could be multiple ways | 21:06 |
dhruv13_ | pirosb3: yeah, like uploads, urls etc right? | 21:06 |
pirosb3 | og | 21:06 |
pirosb3 | oh dhruv13_ do you mean training sets and testing sets? | 21:06 |
dhruv13_ | pirosb3: yeah | 21:07 |
pirosb3 | ok cool | 21:07 |
pirosb3 | unix question: when I use > | 21:07 |
dhruv13_ | pirosb3: i mean, i have no experience with creating web services, i was just wondering how efficient is efficient | 21:07 |
pirosb3 | like "echo hello > file.txt". The write is buffered I mean? | 21:08 |
dhruv13_ | pirosb3: it creates a pipe | 21:08 |
pirosb3 | That would not be a problem dhruv13_, because the rest API would be only used for querying the dataset. The training would all happen "offline" | 21:08 |
pirosb3 | dhruv13_: great. What if we could export some sockets through Docker. Similar to what ZeroMQ does | 21:09 |
dhruv13_ | pirosb3: http://www.cse.ohio-state.edu/~mamrak/CIS762/pipes_lab_notes.html | 21:09 |
pirosb3 | thanks dhruv13_ :) | 21:09 |
pirosb3 | it would be cool to just "write" to the socker | 21:09 |
pirosb3 | socket | 21:09 |
pirosb3 | in a standard format, like svmlight | 21:10 |
dhruv13_ | pirosb3: training offline seriously limits the use cases | 21:11 |
dhruv13_ | pirosb3: provisions should be made, i think | 21:11 |
pirosb3 | but then, wouldnt you always train offline? | 21:11 |
-!- kislay [~kislay@93.114.45.13] has joined #shogun | 21:11 | |
pirosb3 | train offline and serialize | 21:11 |
dhruv13_ | pirosb3: oh, sorry, i meant providing data sets... if it is Saas, then it should take datasets and return a classifier etc. | 21:12 |
pirosb3 | exactly, but all is transparent to the user. which is what makes it super cool | 21:13 |
pirosb3 | (at least thats what I think ahah) | 21:13 |
dhruv13_ | pirosb3: yeah! | 21:13 |
pirosb3 | dhruv13_: what are you thinking of working on? any ideas? :) | 21:13 |
dhruv13_ | pirosb3: the dictionary learning project... I'm reading up on it | 21:13 |
pirosb3 | nice! very cool | 21:14 |
pirosb3 | where are you from dhruv13_ ? | 21:14 |
dhruv13_ | pirosb3: but before that, i'm trying to familiarise myself with shogun codebase | 21:14 |
dhruv13_ | pirosb3: oh, I'm from India, my college is in Goa | 21:14 |
pirosb3 | dhruv13_: same thing over here. I will be working on the codebase today and tomorrow, and work through some examples | 21:15 |
dhruv13_ | pirosb3: :) | 21:15 |
pirosb3 | dhruv13_: Goa must be amazing! at least that's how my indian friend describes it | 21:15 |
dhruv13_ | pirosb3: it's fun, but probably not where you would want to study ;-) | 21:15 |
pirosb3 | ahahah | 21:16 |
pirosb3 | dhruv13_: did you manage to build shogun? | 21:16 |
dhruv13_ | yeah | 21:16 |
@besser82 | dhruv13_: GOA? Ain't that music? | 21:16 |
@besser82 | dhruv13_: :-P | 21:17 |
dhruv13_ | besser82: nope, it's a state in India XD | 21:17 |
dhruv13_ | besser82: famous for beaches and booze | 21:17 |
pirosb3 | dhruv13_: on what OS are you? also, did you install lapack? any other dependencies? | 21:17 |
pirosb3 | besser82: that's exactly what my friend told me! ahah | 21:18 |
dhruv13_ | pirosb3: linux, and yeah, lapack is pretty important | 21:18 |
pirosb3 | dhruv13_: it still will not build for me :( | 21:18 |
@besser82 | dhruv13_: hehe, that's why the music is named like that ;) btw. which linux? | 21:19 |
pirosb3 | are you using develop branch? | 21:19 |
@besser82 | pirosb3: which error? | 21:19 |
dhruv13_ | pirosb3: yeah | 21:19 |
dhruv13_ | pirosb3: this was pretty helpful to me: https://github.com/shogun-toolbox/shogun/blob/develop/doc/md/README_cmake.md | 21:19 |
pirosb3 | still the same, even fater installing lapack /Users/danielpiros/Documents/Projects/shogun/src/shogun/base/class_list.cpp:723:101: error: unknown type name 'CPCA' | 21:19 |
dhruv13_ | besser82: linux mint, but i really don't like it | 21:20 |
dhruv13_ | very buggy | 21:20 |
@besser82 | pirosb3: then it is related to eigen3. did you install that? needs to be >= 3.1.2 | 21:20 |
@besser82 | dhruv13_: wanna try fedora? | 21:20 |
dhruv13_ | besser82: sure | 21:20 |
@besser82 | dhruv13_: there is some nice prebuild rpm for shogun, too ;) | 21:20 |
pirosb3 | yes I did :( I will try with different cmake options | 21:20 |
dhruv13_ | besser82: cool! supports development? | 21:21 |
@besser82 | dhruv13_: by meaning of dependencies, yes ;) | 21:21 |
dhruv13_ | besser82: :) | 21:21 |
@besser82 | dhruv13_: and you can easiely build your own rpm from git-snapshot | 21:22 |
pirosb3 | besser82: should I only build modular interfaces? | 21:22 |
@besser82 | pirosb3: yes, modular interfaces are enough and afaik better supported | 21:22 |
@besser82 | pirosb3: static ones are ancient and left there for legacy | 21:22 |
@besser82 | dhruv13_: are you on broadband or do you need some media? | 21:23 |
dhruv13_ | besser82: I'm very comfortable with debian-like systems... | 21:23 |
dhruv13_ | I'm at home right now.... internet quite slow :-( | 21:24 |
@besser82 | dhruv13_: i thought you want to try it?!? | 21:24 |
dhruv13_ | however, no problem in college... | 21:24 |
dhruv13_ | day after tomorrow? | 21:24 |
@besser82 | dhruv13_: sure, when ever you want to :D | 21:24 |
dhruv13_ | can you point me to the link? | 21:24 |
@besser82 | dhruv13_: if there is trouble with install && stuff, just ping me ---> fedoraproject.org/get-fedora | 21:25 |
dhruv13_ | besser82: sure! | 21:26 |
dhruv13_ | besser82: we have an lug... was going to speak about distros next week | 21:26 |
dhruv13_ | besser82: this might be helpful! :) | 21:26 |
@besser82 | dhruv13_: for sure :D nice to hear | 21:26 |
@besser82 | dhruv13_: fedora is now easy as ubuntu, but has real community ;) | 21:27 |
dhruv13_ | besser82: lol | 21:27 |
dhruv13_ | besser82: are you a dev for them? | 21:28 |
@besser82 | dhruv13_: yes :D | 21:28 |
dhruv13_ | besser82: XD | 21:28 |
@besser82 | dhruv13_: dev, package-sponsor, ambassador, many more | 21:28 |
dhruv13_ | besser82: is it tough to contribute? | 21:28 |
@besser82 | dhruv13_: nope :D Just ask me, I'll point you to right person | 21:28 |
@besser82 | dhruv13_: depending on what you want to contrib | 21:29 |
dhruv13_ | besser82: sure! wanted to learn systems programming at some point | 21:29 |
dhruv13_ | besser82: kernel stuff | 21:29 |
@besser82 | dhruv13_: then fedora is the right place to do so ;) | 21:29 |
dhruv13_ | besser82: be warned, i'm a complete noob! | 21:30 |
@besser82 | dhruv13_: a lot of kernel-devs are from fedora / redhat ;) | 21:30 |
@besser82 | dhruv13_: I believe everybody can learn anything ;) | 21:30 |
dhruv13_ | besser82: sounds fun... | 21:30 |
dhruv13_ | besser82: me too ;-) | 21:30 |
@besser82 | dhruv13_: but for kernel stuff i'm not the right guy to ask ;) | 21:31 |
pirosb3 | besser82: Totally agree "I believe everybody can learn anything " | 21:31 |
@besser82 | dhruv13_: I'm more into userland and ML-stuff | 21:31 |
@besser82 | dhruv13_: and cloud-infra stuffz | 21:31 |
pirosb3 | besser82: where did you start learning ML? | 21:32 |
@besser82 | pirosb3: mostly taught myself | 21:32 |
pirosb3 | yeah same thing I'm trying to do | 21:32 |
@besser82 | pirosb3: but having BSc applied maths helped ;) | 21:32 |
dhruv13_ | besser82: I'm really interested in ML too... but i've not explored kernel dev yet | 21:32 |
pirosb3 | besser82: it does i'm sure! ahahaha | 21:32 |
dhruv13_ | besser82: worked with robots though... | 21:33 |
dhruv13_ | what is userland? | 21:33 |
@besser82 | dhruv13_: me neither ;) Last hardware-related stuff i did was on my C64 | 21:33 |
@besser82 | dhruv13_: commands a user would invoke | 21:33 |
@besser82 | dhruv13_: like some nifty tools and such | 21:34 |
@besser82 | dhruv13_: e.g. jekyll | 21:34 |
dhruv13_ | besser82: cool | 21:34 |
dhruv13_ | besser82: had to google c64 ;-) | 21:34 |
@besser82 | dhruv13_: and system-hardening I forgot ;) | 21:35 |
dhruv13_ | besser82: lol | 21:35 |
@besser82 | dhruv13_: I suppose you are too young to know that ;) | 21:35 |
dhruv13_ | besser82: i wonder how people learnt that stuff | 21:35 |
@besser82 | dhruv13_: which stuff? | 21:35 |
dhruv13_ | besser82: programming before compilers ;-) | 21:36 |
@besser82 | dhruv13_: don't ask :D | 21:36 |
@besser82 | dhruv13_: try to play around with sth. like brainfuck ;) | 21:37 |
dhruv13_ | besser82: i'm all for code optimisation and assembly, but i'm glad i wasn't born earlier XD | 21:37 |
dhruv13_ | ha ha | 21:37 |
@besser82 | dhruv13_: that's how we learned | 21:37 |
@besser82 | dhruv13_: how old are you? | 21:37 |
@besser82 | dhruv13_: I'm 31 | 21:37 |
dhruv13_ | besser82: 21 | 21:37 |
dhruv13_ | besser82: 10 years! | 21:38 |
@besser82 | dhruv13_: yes, the decade where those cool things were invented ! | 21:38 |
dhruv13_ | besser82: ha ha | 21:38 |
@besser82 | dhruv13_: like python or ruby or oop ;) | 21:38 |
dhruv13_ | besser82: ;) | 21:38 |
dhruv13_ | besser82: OOP was there though right? C++? | 21:39 |
dhruv13_ | besser82: no compulsion though... | 21:39 |
@besser82 | dhruv13_: there was some predecessor of c++ | 21:39 |
-!- radgaonkar [~rad@p4FC73B2E.dip0.t-ipconnect.de] has quit [Ping timeout: 265 seconds] | 21:40 | |
@besser82 | dhruv13_: http://en.wikipedia.org/wiki/Simula the first oop lanf | 21:40 |
-!- radgaonkar [~rad@p4FC73B2E.dip0.t-ipconnect.de] has joined #shogun | 21:41 | |
lambday | somebody talked about Goa :D | 21:41 |
dhruv13_ | wow | 21:41 |
dhruv13_ | lambday: I study there! | 21:41 |
dhruv13_ | lambday: I'm from NIT Goa | 21:41 |
lambday | dhruv13_: you're lucky! :D I am from mumbai ;) | 21:42 |
dhruv13_ | lambday: ha ha | 21:42 |
@besser82 | dhruv13_: but C++ was the first oop-lang on "machine level" | 21:42 |
lambday | dhruv13_: well, not from mumbai, but currently studying here | 21:42 |
dhruv13_ | besser82: interesting.... I'm supposed to know this stuff :-p | 21:42 |
sunil1337 | besser82: it was C with classes before c++ i guess | 21:42 |
dhruv13_ | lambday: which college? | 21:43 |
lambday | dhruv13_: iitb | 21:43 |
dhruv13_ | lambday: cool! | 21:43 |
@besser82 | sunil1337: yes, but c with classes was the first draft of c++ :-P | 21:43 |
dhruv13_ | lambday: I'm in mumbai right now, my parents live here | 21:43 |
lambday | dhruv13_: lol :D should have met you for a drink then :P | 21:43 |
@besser82 | sunil1337: it was made by Bjarne Stroustrup as well ;) | 21:44 |
sunil1337 | besser82: :D | 21:44 |
dhruv13_ | lambday: :-D | 21:44 |
dhruv13_ | lambday: which year? 4th? | 21:44 |
lambday | dhruv13_: doing masters - final year | 21:44 |
dhruv13_ | lambday: cool!! | 21:45 |
sunil1337 | besser82: cool! | 21:45 |
lambday | dhruv13_: any chance you know snehalesh/anamay? | 21:45 |
dhruv13_ | lambday: nope, sorry... | 21:46 |
dhruv13_ | lambday: I'll probably know people from pace | 21:46 |
dhruv13_ | lambday: or from JK Singhania School | 21:46 |
dhruv13_ | lambday: in Thane | 21:46 |
lambday | dhruv13_: lol I don't know which schools they are from | 21:46 |
dhruv13_ | lambday: XD okay! | 21:47 |
sunil1337 | besser82: do u knw where the python modules are located after they're built? | 21:47 |
lambday | dhruv13_: how's your shogun stuff going? | 21:48 |
dhruv13_ | lambday: submitted my first PR | 21:48 |
dhruv13_ | lambday: issue #1888 | 21:48 |
@besser82 | sunil1337: yes, build-dir/shogun/interfaces/*_modular/ | 21:48 |
lambday | dhruv13_: buy yourself a drink when it gets merged - that's what I did - great feeling :d | 21:49 |
dhruv13_ | lambday: https://github.com/shogun-toolbox/shogun/pull/1896 | 21:49 |
dhruv13_ | lambday: hope it's soon! | 21:49 |
sunil1337 | besser82: there's no sg.so in there... | 21:49 |
@besser82 | sunil1337: possibly there may be another underneath sub-dir with the so-plugin.. | 21:49 |
@besser82 | sunil1337: ^^ | 21:49 |
dhruv13_ | lambday: sure! | 21:49 |
lambday | dhruv13_: good to see you man! carry on the good work :) | 21:50 |
dhruv13_ | lambday: thanks!! same here!! | 21:50 |
@besser82 | sunil1337: no, but some _modshogun.so ;) | 21:50 |
@besser82 | sunil1337: like in python_modular | 21:50 |
pirosb3 | nice pull request dhruv13_ ! | 21:50 |
sunil1337 | besser82: ya i have that... but python fails to import every after specifying the lib dir. | 21:51 |
sunil1337 | besser82: even* | 21:51 |
@besser82 | sunil1337: libdir? | 21:51 |
@besser82 | sunil1337: try PATHON_PATH ;) | 21:51 |
sunil1337 | besser82: i meant the path. | 21:51 |
@besser82 | sunil1337: try PYTHON_PATH ;) | 21:51 |
@besser82 | sunil1337: a ok | 21:51 |
dhruv13_ | pirosb3: thanks! but failed travis ci many times... | 21:52 |
@besser82 | sunil1337: which failes.... | 21:52 |
@besser82 | sunil1337: to import modshogun? | 21:52 |
sunil1337 | besser82: yup | 21:52 |
dhruv13_ | pirosb3: tests passed locally, should be okay this time round... | 21:52 |
@besser82 | sunil1337: mhhh.... just a moment | 21:52 |
lambday | besser82: in fedora its PYTHONPATH, no? (without the underscore) | 21:52 |
pirosb3 | ok | 21:52 |
dhruv13_ | besser82: PYTHONPATH by default on most linux systems i think... | 21:53 |
@besser82 | lambday: yes, sry my fault... | 21:53 |
@besser82 | sunil1337: what is the excat error msg? | 21:54 |
sunil1337 | besser82: no module named modshogun found | 21:54 |
sunil1337 | besser82: i had tried python_static | 21:55 |
sunil1337 | besser82: i'm recompiling with python_modules lets see if this works.. | 21:55 |
sunil1337 | besser82: pythonmodular* | 21:55 |
@besser82 | sunil1337: yes, try modular ;) It's better supported ;) | 21:56 |
sunil1337 | besser82: hmm...sure thanks.. | 21:56 |
sunil1337 | besser82: where r u from? | 21:56 |
@besser82 | sunil1337: from Germany, near Hannover | 21:56 |
@besser82 | sunil1337: and you? | 21:56 |
sunil1337 | besser82: india from hyderabad | 21:57 |
@besser82 | sunil1337: sounds nice | 21:59 |
dhruv13_ | besser82: what's the role of ML in fedora? | 21:59 |
@besser82 | sunil1337: so it already must be past midnight @ yours, right?? | 21:59 |
sunil1337 | besser82: yup... its 2:30 to be exact | 21:59 |
dhruv13_ | besser82: 2:30 A.M! | 21:59 |
dhruv13_ | sunil1337: lol | 22:00 |
@besser82 | dhruv13_: currently not so strong, but we are growing ;) | 22:00 |
lambday | besser82: city of nawabs (kings, sort of) - with awesome dishes, called "biryani" :D | 22:00 |
sunil1337 | dhruv13_: nitians bro :p | 22:00 |
dhruv13_ | sunil1337: cool! warangal? | 22:00 |
@besser82 | lambday: biryani? Heard of them somewhen before... | 22:00 |
sunil1337 | dhruv13_: village rourkela :D | 22:00 |
dhruv13_ | sunil1337: lol | 22:00 |
@besser82 | lambday: but what is that exactly? | 22:01 |
lambday | besser82: yeah that's a special mughlai dish - you should try it sometime... | 22:01 |
dhruv13_ | sunil1337: goa is pretty new... in fact i'm part of the first batch!! | 22:01 |
dhruv13_ | besser82: it is spiced rice | 22:01 |
lambday | besser82: well, its rice with spices and chicken/mutton/beef - you should definitely try it out in some Indian restaurant | 22:01 |
dhruv13_ | besser82: though that is selling it too shortly ;-) | 22:01 |
sunil1337 | dhruv13_: cool! u establish everything there.. the culture and stuff right? | 22:01 |
dhruv13_ | sunil1337: ha ha, sort of | 22:02 |
sunil1337 | http://upload.wikimedia.org/wikipedia/commons/f/fe/Chicken_Biryani.jpg if u r feeling hungry :p | 22:02 |
lambday | besser82: btw you weren't talking about goa trance music, were you? | 22:02 |
@besser82 | lambday, dhruv13_: sound quite tasty :D Will try it, when I'm next time in "Göttingen" (another city nearby) @ me fav. indian-restaurant | 22:03 |
@besser82 | lambday: yes, I did :D | 22:03 |
@besser82 | lambday: Trance rockzzzzz | 22:03 |
lambday | besser82: Armin van Burren :D | 22:03 |
@besser82 | lambday: nice! | 22:03 |
lambday | besser82: hey btw shogun is a trace dj's name :d | 22:03 |
sunil1337 | dhruv13_: hows food there bro? | 22:03 |
@besser82 | lambday: never heard of him... | 22:04 |
lambday | besser82: wait a sec :P | 22:04 |
lambday | besser82: http://www.youtube.com/watch?v=3x0Xdy5kGoY | 22:04 |
dhruv13_ | sunil1337: in goa, great if you like sea food... pretty limited if you are vegetarian | 22:05 |
dhruv13_ | sunil1337: hostel mess, well what do you expect ;-P | 22:05 |
sunil1337 | dhruv13_: it sucks here :/ | 22:05 |
lambday | dhruv13_: lobsterrssssss :D kingfish!!!! | 22:05 |
dhruv13_ | sunil1337: i can imagine! | 22:05 |
lambday | fuck I am feeling hungry | 22:05 |
sunil1337 | lambday: i'm always hungry :p | 22:06 |
dhruv13_ | me too! | 22:06 |
lambday | sunil1337: lol D: | 22:06 |
lambday | besser82: if you ever to visit India, never miss Goa! | 22:06 |
dhruv13_ | luckily guys, I'm at home!! | 22:06 |
dhruv13_ | besser82: yeah! | 22:06 |
@besser82 | lambday: I'll take a visit for sure !!! dhruv13_ | 22:06 |
sunil1337 | dhruv13_: do u have any idea when our ends would be? | 22:06 |
@besser82 | lambday: btw. let's implement those sounds into shogun XD | 22:07 |
dhruv13_ | boy, we're really mucking up the irc aren't we?! | 22:07 |
-!- votjakovr [~votjakovr@188.134.46.30] has quit [Quit: WeeChat 0.4.0] | 22:07 | |
sunil1337 | dhruv13_: ya :p | 22:07 |
lambday | besser82: lol..! count me in :D | 22:07 |
-!- pirosb3_ [8da3303c@gateway/web/freenode/ip.141.163.48.60] has joined #shogun | 22:07 | |
@besser82 | dhruv13_: nope not really ;) | 22:07 |
pirosb3_ | Hey guys, I'll be going! | 22:07 |
dhruv13_ | pirosb3: bye! | 22:07 |
pirosb3_ | hear you soon! :) | 22:07 |
dhruv13_ | sure! | 22:08 |
sunil1337 | besser82: stuck at 86% with pythonmodular :/ | 22:08 |
@besser82 | pirosb3_: good night! | 22:08 |
lambday | pirosb3_: didn't get a chance to get introduced - see you :) | 22:08 |
dhruv13_ | pirosb3: good night! | 22:08 |
@besser82 | sunil1337: ??? | 22:08 |
@besser82 | sunil1337: compiler error or what? | 22:08 |
pirosb3_ | Oh don't worry. I think we will be hearing each other in the next days :) | 22:08 |
sunil1337 | besser82: just stuck. | 22:08 |
pirosb3_ | see ya! | 22:09 |
lambday | pirosb3_: ;) ciao man! | 22:09 |
@besser82 | sunil1337: don't worry. building those large (15MB+) swig-generated c++-code takes time ;) | 22:09 |
-!- pirosb3 [8da3303c@gateway/web/freenode/ip.141.163.48.60] has quit [Ping timeout: 245 seconds] | 22:09 | |
lambday | sunil1337: yeah it takes a while :( | 22:09 |
sunil1337 | besser82: stuck with warnings... okay! will wait | 22:09 |
@besser82 | sunil1337: linking takes another piece of time as well | 22:10 |
@besser82 | sunil1337: which warnings? | 22:10 |
dhruv13_ | c++ flags deprecated? | 22:10 |
sunil1337 | besser82: using numpy depricated | 22:10 |
@besser82 | sunil1337: you can ignore that ;) | 22:10 |
sunil1337 | besser82: hmm... :) | 22:11 |
@besser82 | sunil1337: that will need fixing as soon as numpy 1.8.0 has hit even those real rock-solid distros like RHEL ;) | 22:11 |
@besser82 | sunil1337: it just tells you to update your code in future somewhen ;) | 22:12 |
sunil1337 | besser82: okay...let me get over with this... | 22:12 |
-!- pirosb3_ [8da3303c@gateway/web/freenode/ip.141.163.48.60] has quit [Ping timeout: 245 seconds] | 22:13 | |
-!- tomy18 [~alex@p6.eregie.pub.ro] has quit [Ping timeout: 260 seconds] | 22:13 | |
lambday | sonney2k: why index_t is not uint32_t and int32_t? | 22:17 |
lambday | but* | 22:17 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * 88335ed / src/interfaces/modular/modshogun.i: https://github.com/shogun-toolbox/shogun/commit/88335ed55ece17f42452c6dc9e6072f3ad9dfb32 | 22:19 |
shogun-notifier- | shogun: drop the package name so we will have names modshogun.CLASS with swig 3.0 | 22:19 |
-!- pickle27 [c0008876@gateway/web/freenode/ip.192.0.136.118] has quit [Quit: Page closed] | 22:24 | |
sunil1337 | bye guys i'm off. | 22:30 |
-!- sunil1337 [0e8bd057@gateway/web/freenode/ip.14.139.208.87] has quit [Quit: Page closed] | 22:32 | |
-!- lisitsyn [~lisitsyn@213.87.131.123] has joined #shogun | 22:32 | |
lisitsyn | whoaa | 22:32 |
@HeikoS | reeee | 22:33 |
@HeikoS | lisitsyn: woaaaa! | 22:33 |
lisitsyn | HeikoS: I am back! | 22:33 |
@HeikoS | interview groupies? :) | 22:34 |
@besser82 | lisitsyn: hey, man! 'sup? | 22:38 |
lisitsyn | besser82: hey | 22:39 |
lisitsyn | HeikoS: it went rather ok I think ;) | 22:39 |
* besser82 needs some Zzzzzzz now, too, he thinks ;) | 22:40 | |
@besser82 | see you tomorrow, guys! :D | 22:40 |
lisitsyn | besser82: see ya! | 22:40 |
@besser82 | lisitsyn: r u on tommorow? | 22:41 |
dhruv13_ | besser82: see you! good night!! | 22:41 |
@besser82 | lisitsyn: /back | 22:41 |
lisitsyn | besser82: what is going to happen tomorrow? ;) | 22:41 |
@besser82 | l/back | 22:41 |
@besser82 | lisitsyn: nothing speciuhl ;) | 22:42 |
lisitsyn | I just got back from the airport :) | 22:42 |
@besser82 | lisitsyn: perhaps CCCP 2.0 occupies some more countries ;) | 22:42 |
lisitsyn | besser82: fear that! | 22:42 |
@besser82 | lisitsyn: I'm praying everyday not have that happening | 22:43 |
@besser82 | lisitsyn: I was just asking to have some chit-chat chan-spam with you tomorrow :-P | 22:44 |
@besser82 | lisitsyn: now I'm really off for today... | 22:44 |
@HeikoS | dhruv13_: bye! | 22:44 |
dhruv13_ | HeikoS: bye! good night!! | 22:44 |
@HeikoS | lisitsyn: so, tell us about it | 22:44 |
@HeikoS | lisitsyn: or jus tme proivately :) | 22:44 |
dhruv13_ | HeikoS: wait, i want to discuss https://github.com/shogun-toolbox/shogun/issues/1918 | 22:46 |
@HeikoS | dhruv13_: ok | 22:46 |
dhruv13_ | HeikoS: I'm referring to "A kernel statistical test of independence" paper for the HSIC class | 22:47 |
@HeikoS | dhruv13_: so just to let you know, lambday is currently working on these classes, though he focusses on different things, but there will be some major changes in the near future | 22:47 |
dhruv13_ | what changes do you need to be done? | 22:47 |
dhruv13_ | oh | 22:47 |
@HeikoS | dhruv13_: yeah thats the one | 22:47 |
dhruv13_ | lambday: can I help? | 22:47 |
@HeikoS | dhruv13_: so you could let lambday know that you are working on this one, then no conflicts | 22:47 |
@HeikoS | dhruv13_, lambday you two want to make sure to not work on the same thing | 22:47 |
dhruv13_ | HeikoS: okay :) | 22:47 |
@HeikoS | dhruv13_: so for the number of points to be different, a couple of things need to be changed | 22:48 |
@HeikoS | dhruv13_: currently, all data is stored in one feature object, and there is a variable that says from which index the second dataset starts | 22:48 |
lambday | dhruv13_: I am currently changing a couple of other thing - thought of turning to that issue after these | 22:48 |
dhruv13_ | lambday: I'll read up and see if i can grasp what changes i can make | 22:49 |
lambday | HeikoS: yeah but I just changed all that :-/ | 22:49 |
@HeikoS | dhruv13_: this has to allow to be different than in the middle, and then computing the statistic is the main thing that changes | 22:49 |
dhruv13_ | lambday: oh, is it done already? I'll look at a different issue then | 22:49 |
@HeikoS | dhruv13_: and the spectral test for null approximation | 22:49 |
@HeikoS | dhruv13_: yeah lambday is changing class structure, so maybe go for something else for now | 22:50 |
dhruv13_ | HeikoS, lambday : okay :) any suggestions? | 22:50 |
@HeikoS | dhruv13_: I mean there are more issues than anyone can solve this year :D | 22:50 |
dhruv13_ | sure! | 22:51 |
dhruv13_ | HeikoS: python integration test failed again :-(... SerializationJSON* passed though... | 22:58 |
@HeikoS | dhruv13_: let me check | 22:58 |
dhruv13_ | HeikoS: I'm not sure where the equals method of TParameter is being called in the file... | 23:00 |
@HeikoS | dhruv13_: just a sec checking | 23:01 |
@HeikoS | dhruv13_: just got a call, will take 5 mins | 23:01 |
dhruv13_ | HeikoS: no problem! | 23:01 |
@HeikoS | back | 23:04 |
shogun-notifier- | shogun: Parijat Mazumdar :develop * 5cae1ab / src/shogun/preprocessor/PCA.cpp,tests/unit/preprocessor/PCA_unittest.cc: https://github.com/shogun-toolbox/shogun/commit/5cae1ab5c9adcb6fdd0b4b79573477f743776114 | 23:05 |
shogun-notifier- | shogun: PCA eigen-ized | 23:05 |
shogun-notifier- | shogun: Parijat Mazumdar :develop * 51adc46 / / (3 files): https://github.com/shogun-toolbox/shogun/commit/51adc460daf37ac808bf1a2ff31033dc9f73e065 | 23:05 |
@HeikoS | first | 23:05 |
shogun-notifier- | shogun: SVD based PCA added | 23:05 |
shogun-notifier- | shogun: Parijat Mazumdar :develop * fe1cbeb / src/shogun/preprocessor/PCA.cpp: https://github.com/shogun-toolbox/shogun/commit/fe1cbeb4a9ed5f44c2775330b489c513120d2cf3 | 23:05 |
shogun-notifier- | shogun: code cleaned | 23:05 |
shogun-notifier- | shogun: Parijat Mazumdar :develop * e737c4e / tests/unit/metric/LMNN_unittest.cc: https://github.com/shogun-toolbox/shogun/commit/e737c4e7328ca7b1aaf0d79a3c42a640ba8e3fb1 | 23:05 |
shogun-notifier- | shogun: failing LMNN unit-test : test_pca_init changed | 23:05 |
shogun-notifier- | shogun: Parijat Mazumdar :develop * 2e5573d / / (4 files): https://github.com/shogun-toolbox/shogun/commit/2e5573d9ddb724c1920ce4d06b8065385c69e83e | 23:05 |
shogun-notifier- | shogun: brief documentation updated + apply_to_feature_vector updated | 23:05 |
shogun-notifier- | shogun: Parijat Mazumdar :develop * 6644ff3 / / (3 files): https://github.com/shogun-toolbox/shogun/commit/6644ff31e15e3b7a00a2e3dc0b9e100e5e396feb | 23:05 |
shogun-notifier- | shogun: PCA code cleanup | 23:05 |
shogun-notifier- | shogun: Parijat Mazumdar :develop * 438d455 / data: https://github.com/shogun-toolbox/shogun/commit/438d455c1226244684e748ff262e197442822c07 | 23:05 |
shogun-notifier- | shogun: updating revision of data submodule | 23:05 |
shogun-notifier- | shogun: Parijat Mazumdar :develop * 58d5f32 / data: https://github.com/shogun-toolbox/shogun/commit/58d5f3243456651d6bf0434834d3771693420c65 | 23:05 |
shogun-notifier- | shogun: update data submodule : lmnn_modular changed | 23:05 |
shogun-notifier- | shogun: Parijat Mazumdar :develop * d4f7a65 / src/shogun/ (3 files): https://github.com/shogun-toolbox/shogun/commit/d4f7a65d7f4746d3441a13ea2d5209a79ac182bc | 23:05 |
shogun-notifier- | shogun: HAVE_LAPACK to HAVE_EIGEN3 and header changes | 23:05 |
@HeikoS | dhruv13_: ok so kernelpca modular fails | 23:07 |
dhruv13_ | HeikoS: yeah | 23:07 |
dhruv13_ | HeikoS: didn't have that problem when i tested locally | 23:08 |
dhruv13_ | HeikoS: or probably didn't compile integration tests | 23:08 |
dhruv13_ | HeikoS: i think the latter | 23:08 |
@HeikoS | do you compile with python modular? | 23:08 |
dhruv13_ | HeikoS: yes | 23:08 |
dhruv13_ | HeikoS: what script calls kernelpca? I just have to set tolerant to true in the equals call | 23:09 |
@HeikoS | what happens if you run tester.py in the python modular integration testing dir? | 23:09 |
dhruv13_ | the error is similar to JSON Serialization | 23:09 |
dhruv13_ | HeikoS: wait, i'll try | 23:10 |
@HeikoS | dhruv13_: yeah, its a small negative number | 23:10 |
@HeikoS | dhruv13_: btw the error output is weird | 23:10 |
@HeikoS | it only tells usleaving TParameter::compare_ptype(): | 23:10 |
@HeikoS | leaving TParameter::compare_ptype(): | 23:10 |
shogun-buildbot | build #2723 of deb1 - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun/builds/2723 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Parijat Mazumdar <mazumdarparijat@gmail.com> | 23:10 |
@HeikoS | and no entering | 23:10 |
dhruv13_ | HeikoS: yeah... | 23:11 |
dhruv13_ | HeikoS: is that because of the --output-on-failure | 23:11 |
dhruv13_ | ? | 23:11 |
@wiking | shogun-buildbot: force build --branch=develop 'deb1 - libshogun' | 23:12 |
shogun-buildbot | build forced [ETA 2m20s] | 23:12 |
shogun-buildbot | I'll give a shout when the build finishes | 23:12 |
@HeikoS | dhruv13_: I think its because wrong method to print stuff | 23:13 |
@HeikoS | one is SG_INFO the others are debug or so | 23:13 |
@HeikoS | dhruv13_: but dont worry about that for now, we want to fix the error | 23:13 |
dhruv13_ | HeikoS: | 23:13 |
dhruv13_ | HeikoS: yeah | 23:13 |
dhruv13_ | HeikoS: which script calls kernelpca* for the integration test? | 23:14 |
@HeikoS | run tester.py kernelpca_modular (pr whatever the name is) | 23:14 |
shogun-buildbot | build #2724 of deb1 - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun/builds/2724 | 23:14 |
dhruv13_ | it's not built... i ran cmake with python_modular on... | 23:15 |
dhruv13_ | strange | 23:15 |
@HeikoS | tester.py is in tests/integratoin/python_modular | 23:15 |
dhruv13_ | not necessary to build? | 23:15 |
@HeikoS | wiking: thats the thing with this MLData failure on the buildbot? | 23:15 |
@HeikoS | dhruv13_: no these are python examples | 23:15 |
@HeikoS | dhruv13_: as long as you compiled the binding you can run any python script that interfaces shogun | 23:16 |
dhruv13_ | okay... i thought the calling script would be c++ or something | 23:16 |
@HeikoS | nono | 23:16 |
thoralf | lol | 23:16 |
thoralf | Who uses C++? ;) | 23:16 |
dhruv13_ | thoralf: yeah! | 23:16 |
@wiking | HeikoS: noidea.. mldata has some problems sometimes ;P | 23:16 |
dhruv13_ | thoralf: most i have used in years, these past 4 days!! | 23:17 |
@HeikoS | thoralf: haha ;) | 23:17 |
thoralf | dhruv13_: No worries. It's like Java, but makes bigger holes if you miss your target. ;) | 23:17 |
@HeikoS | dhruv13_, thoralf yeah it can be a little bitchy that language | 23:18 |
dhruv13_ | HeikoS: lol | 23:18 |
dhruv13_ | tester.py complains that there are no data files | 23:22 |
@HeikoS | dhruv13_: did you do git submodule init? | 23:22 |
@HeikoS | and git submodule update? | 23:22 |
@HeikoS | (which checks out the data repository which contains the integration testing files= | 23:22 |
dhruv13_ | ? no | 23:23 |
dhruv13_ | okay, on it's way :) | 23:24 |
dhruv13_ | might take a while... internet's really slow | 23:24 |
@HeikoS | dhruv13_: yep its big! | 23:24 |
dhruv13_ | 60-80 kbps | 23:24 |
dhruv13_ | how big? | 23:24 |
dhruv13_ | HeikoS: i just need to know which is the glue script | 23:25 |
dhruv13_ | i mean, the one which compares the output of tests to the stored ones | 23:26 |
dhruv13_ | there seems to be a truncated value stored, which is why it seems to be failing | 23:26 |
dhruv13_ | but yeah, i'll be patient and test locally before my next pr :-P | 23:27 |
@HeikoS | dhruv13_: yeah thats important to keep travis not too busy | 23:27 |
@HeikoS | dhruv13_: since we got so many PR, it sometimes takes a while to build, so local checks are good! :) | 23:27 |
dhruv13_ | HeikoS: I've kept it quite busy ;-) | 23:28 |
kislay | I am working on issue #1878. I have to complete all the ideas written there or some will do? | 23:34 |
@HeikoS | kislay: No, some ideas will do, but you should aim for high quality here. Rather focus on just one method (say PCA) and try to write a notebook as nice as possible for that. We can then extend later. | 23:36 |
kislay | I managed to complete the a visual 'eye candy' sort of thing on 2d and 3d data | 23:36 |
@HeikoS | kislay: and see the latest merged patch which changed PCA quite a bit | 23:36 |
@HeikoS | kislay: thats great, feel free to share this | 23:36 |
kislay | hmm... I love eigenfaces. letme add that there. It will be fun :) | 23:37 |
@HeikoS | kislay: that would be *great* because noone has nice code for that | 23:37 |
kislay | I once implemented in matlab. | 23:38 |
kislay | hope that will help me. | 23:38 |
@HeikoS | kislay: I am sure it will | 23:38 |
@HeikoS | kislay: see also #1915 | 23:38 |
kislay | :) | 23:38 |
@HeikoS | we just had a massive update on PCA | 23:38 |
kislay | Oh! | 23:39 |
kislay | one thing! will the old implementation of PCA abandoned...or it's still compatible? | 23:44 |
@HeikoS | kislay: no we kept everything compatible, all examples should still work | 23:44 |
@HeikoS | just watch out for the new fetaures | 23:44 |
kislay | sure sure | 23:44 |
@HeikoS | wiking: how is the notebook slave going? :) | 23:45 |
@HeikoS | kislay: best thing is to pull, rebuild, and then see if your stuff still works | 23:45 |
kislay | yup! | 23:45 |
kislay | I must sleep! 4:18 AM already. good night people!! :) | 23:49 |
@HeikoS | kislay: haha, good stuff :) | 23:49 |
@HeikoS | kislay: sleep well! | 23:49 |
dhruv13_ | kislay: good night! ISThuh? | 23:50 |
dhruv13_ | kislay: me too by the way ! | 23:50 |
@HeikoS | dhruv13_: haha, crazy guys | 23:52 |
@HeikoS | dhruv13_: any news on the integration tests? | 23:52 |
dhruv13_ | 131 mb done zzz | 23:52 |
@HeikoS | dhruv13_: ah man thats slow | 23:52 |
@HeikoS | dhruv13_: let me check the size | 23:52 |
dhruv13_ | i'm trying to figure out where the call to equals is | 23:53 |
@HeikoS | dhruv13_: 366 :/ | 23:53 |
dhruv13_ | will be another hour at least :/ | 23:53 |
-!- kislay [~kislay@93.114.45.13] has quit [Ping timeout: 240 seconds] | 23:54 | |
thoralf | dhruv13_: It's like in good-old modem days. :) | 23:56 |
dhruv13_ | thoralf: tell me about it :-( | 23:57 |
thoralf | 56kbit/s, waiting the whole night for the por^Wprogramm download to finish... | 23:57 |
dhruv13_ | ha ha | 23:58 |
thoralf | That's when I learned "never underestimate a truckload full of tapes" -- and it's still true today. | 23:58 |
dhruv13_ | yeah | 23:58 |
dhruv13_ | tanenbaum!!! | 23:59 |
thoralf | Yes? | 23:59 |
dhruv13_ | the quote: "Never underestimate a stagecoach full of tapes hurtling down the highway" | 23:59 |
dhruv13_ | in his networks book | 23:59 |
--- Log closed Tue Mar 04 00:00:06 2014 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!