--- Log opened Sun Mar 20 00:00:38 2016 | ||
-!- timsl [54d8e30b@gateway/web/freenode/ip.84.216.227.11] has quit [Quit: Page closed] | 00:26 | |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 01:29 | |
shogun-buildbot | build #974 of nightly_all is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/nightly_all/builds/974 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, trevor <trevor.ballard@outlook.com>, Wu Lin <yorker.lin@gmail.com> | 03:06 |
---|---|---|
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun | 03:09 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 03:09 | |
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has quit [Client Quit] | 03:11 | |
shogun-buildbot | build #1101 of nightly_default is complete: Failure [failed test notebooks] Build details are at http://buildbot.shogun-toolbox.org/builders/nightly_default/builds/1101 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, trevor <trevor.ballard@outlook.com>, Wu Lin <yorker.lin@gmail.com> | 03:25 |
-!- lambday [6a33e9dd@gateway/web/freenode/ip.106.51.233.221] has joined #shogun | 06:10 | |
-!- mode/#shogun [+o lambday] by ChanServ | 06:10 | |
-!- sonne|osx [~sonne@x4db4d6d7.dyn.telefonica.de] has joined #shogun | 06:13 | |
-!- lambday [6a33e9dd@gateway/web/freenode/ip.106.51.233.221] has quit [Ping timeout: 252 seconds] | 06:25 | |
-!- sonne|osx [~sonne@x4db4d6d7.dyn.telefonica.de] has quit [Quit: sonne|osx] | 06:45 | |
-!- lambday [6a33f351@gateway/web/freenode/ip.106.51.243.81] has joined #shogun | 07:50 | |
-!- mode/#shogun [+o lambday] by ChanServ | 07:50 | |
-!- besser82 [~besser82@fedora/besser82] has joined #shogun | 09:33 | |
-!- mode/#shogun [+o besser82] by ChanServ | 09:33 | |
-!- curiousguy13 [~curiousgu@59.177.196.139] has joined #shogun | 10:00 | |
-!- curiousguy13 [~curiousgu@59.177.196.139] has quit [Ping timeout: 260 seconds] | 10:45 | |
-!- sshkhr [7d1409e6@gateway/web/freenode/ip.125.20.9.230] has joined #shogun | 10:51 | |
-!- curiousguy13 [~curiousgu@120.57.233.223] has joined #shogun | 10:58 | |
sshkhr | I have a small confusion in the Install wiki instructions. Anyone online to help me out ? | 11:04 |
-!- sanuj [0e8bc402@gateway/web/freenode/ip.14.139.196.2] has joined #shogun | 11:18 | |
sanuj | lambday: there? | 11:21 |
-!- besser82 [~besser82@fedora/besser82] has quit [Ping timeout: 240 seconds] | 11:32 | |
-!- mizari [~mizari@95-174-213-100.nts.su] has quit [Read error: Connection reset by peer] | 11:48 | |
-!- mizari [~mizari@95-174-213-100.nts.su] has joined #shogun | 12:07 | |
-!- lambday [6a33f351@gateway/web/freenode/ip.106.51.243.81] has quit [Ping timeout: 252 seconds] | 12:11 | |
-!- sanuj [0e8bc402@gateway/web/freenode/ip.14.139.196.2] has quit [Ping timeout: 252 seconds] | 12:17 | |
-!- sonne|osx [~sonne@x4db4d6d7.dyn.telefonica.de] has joined #shogun | 13:02 | |
-!- sonne|osx [~sonne@x4db4d6d7.dyn.telefonica.de] has quit [Quit: sonne|osx] | 13:24 | |
-!- Nebula_ [~chatzilla@1.39.63.29] has joined #shogun | 14:03 | |
-!- Nebula_ [~chatzilla@1.39.63.29] has left #shogun [] | 14:13 | |
-!- sshkhr [7d1409e6@gateway/web/freenode/ip.125.20.9.230] has quit [Quit: Page closed] | 14:17 | |
-!- lambday [6a33f351@gateway/web/freenode/ip.106.51.243.81] has joined #shogun | 14:35 | |
-!- mode/#shogun [+o lambday] by ChanServ | 14:35 | |
-!- sanuj [0e8bc402@gateway/web/freenode/ip.14.139.196.2] has joined #shogun | 15:05 | |
-!- mizari [~mizari@95-174-213-100.nts.su] has quit [Ping timeout: 276 seconds] | 16:04 | |
-!- mizari [~mizari@95-174-213-100.nts.su] has joined #shogun | 16:10 | |
-!- livus [~suvil@196.207.84.20] has joined #shogun | 16:12 | |
-!- mizari [~mizari@95-174-213-100.nts.su] has quit [Ping timeout: 268 seconds] | 16:28 | |
-!- mizari [~mizari@95-174-213-100.nts.su] has joined #shogun | 16:32 | |
-!- sanuj [0e8bc402@gateway/web/freenode/ip.14.139.196.2] has quit [Quit: Page closed] | 16:48 | |
-!- besser82 [~besser82@fedora/besser82] has joined #shogun | 17:30 | |
-!- mode/#shogun [+o besser82] by ChanServ | 17:30 | |
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun | 17:51 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 17:51 | |
-!- witness_ [uid10044@gateway/web/irccloud.com/x-lxxoegdxpfzkglbv] has joined #shogun | 17:55 | |
-!- livus [~suvil@196.207.84.20] has quit [Ping timeout: 240 seconds] | 17:57 | |
@lambday | HeikoS: there? | 17:58 |
@HeikoS | lambday: jojo | 17:58 |
@lambday | jojo | 17:58 |
@lambday | HeikoS: https://github.com/lambday/flash/blob/develop/tests/MMDTest.cpp | 17:59 |
@lambday | MMD more or less complete | 17:59 |
@lambday | HeikoS: will feature branch it from tomorrow | 17:59 |
@HeikoS | checking | 18:00 |
@HeikoS | compute_variance computes variance of the estimate? | 18:00 |
@HeikoS | that is, variance of the h-terms? | 18:00 |
@HeikoS | where mmd itself is mean of the h-terms? | 18:01 |
@lambday | HeikoS: no it is variance estimation under null.. | 18:01 |
-!- livus [~suvil@196.207.84.20] has joined #shogun | 18:01 | |
@lambday | HeikoS: it uses the variance of h-terms when the estimation method is PERMUTATION.. otherwise, WITHIN_BLOCK_DIRECT is somewhat different | 18:02 |
@HeikoS | lambday: cool looks like all interfaces seem to wrk? | 18:02 |
@HeikoS | ok I see | 18:02 |
@lambday | HeikoS: almost | 18:02 |
@HeikoS | lambday: is this stuff running on multiple threads already? | 18:02 |
@lambday | HeikoS: yep! | 18:02 |
@HeikoS | lambday: cool stuff :) | 18:03 |
@HeikoS | lambday: looks very clean to use the interface | 18:03 |
@lambday | HeikoS: will add gpu stuff when it's inside shogun | 18:03 |
@HeikoS | lambday: yeah later is fine for that | 18:03 |
@HeikoS | lambday: cool so I can start putting examples in soon | 18:03 |
@lambday | HeikoS: API-wise it is pretty similar.. but the internals are quite different | 18:03 |
@HeikoS | yeah | 18:04 |
@HeikoS | cool | 18:04 |
@HeikoS | I will do some high level experimental sanity checks once all this is in feature branch | 18:04 |
@HeikoS | on the base of the existing notebook | 18:04 |
@lambday | HeikoS: absolutely.. | 18:05 |
@lambday | HeikoS: I was thinking of adding all of these as it is inside the feature branch.. so that we can start writing API examples.. then I'll iteratively add proper licensing, documentation and unit-tests | 18:06 |
@HeikoS | lambday: yes | 18:06 |
@HeikoS | good | 18:06 |
@HeikoS | lambday: then we can parallelise the work a bit | 18:06 |
@lambday | yeah that's what I was thinking | 18:06 |
@HeikoS | lambday: I will first write cookbook examples for all basic cases | 18:07 |
@HeikoS | lambday: and then do some simple statistical tests to make sure these things behave as expected | 18:08 |
lisitsyn | hey | 18:08 |
@lambday | HeikoS: yeah that sounds like a plan | 18:08 |
@lambday | HeikoS: BTW I'll be going home next week on Wednesday.. will come back on Tuesday.. hopefully I can still work while I'm at home.. | 18:09 |
@lambday | lisitsyn: hola senor | 18:09 |
-!- sanuj [0e8bc402@gateway/web/freenode/ip.14.139.196.2] has joined #shogun | 18:09 | |
@HeikoS | lisitsyn: yoz! | 18:09 |
@HeikoS | lambday: ok cool | 18:09 |
lisitsyn | HeikoS: git-deploy is on hold as I need some help from wiking | 18:09 |
lisitsyn | (no rights, as usual) | 18:10 |
@HeikoS | lambday: good if the feature branch was running before you left ... so I can play | 18:10 |
@HeikoS | lisitsyn: ah man :) | 18:10 |
@lambday | HeikoS: BTW I created an issue for linalg that we discussed.. would start working on that as well | 18:10 |
@HeikoS | lambday: cool | 18:10 |
@HeikoS | yeah just checking emails | 18:10 |
@lambday | HeikoS: yes I'll do most of it by today/tomorrow | 18:10 |
@HeikoS | lambday: maybe create an entrance task if there are simple bits | 18:10 |
@lambday | HeikoS: first few bytes I'll write.. entrance tasks shouldn't be this much confusing :D I'll write once I refactor some of it, get the skeleton running | 18:11 |
@HeikoS | lambday: I guess thats a good point | 18:12 |
@HeikoS | lisitsyn: did you ahve a look on this? | 18:12 |
lisitsyn | HeikoS: this = ? | 18:12 |
@HeikoS | lisitsyn: the linalg updates lambday thought of | 18:13 |
@lambday | lisitsyn: I guess he was talking about https://github.com/shogun-toolbox/shogun/issues/3100 | 18:13 |
@HeikoS | lisitsyn: more clean distinction between library and hardware backend | 18:13 |
lisitsyn | wow that's big | 18:13 |
@lambday | hahaha | 18:13 |
@lambday | read it, it's funny | 18:14 |
@HeikoS | lambday: haha | 18:14 |
@lambday | specially if you | 18:14 |
@lambday | 're following news | 18:14 |
@HeikoS | just reading it | 18:14 |
@HeikoS | haha | 18:14 |
@HeikoS | deep drump ;) | 18:14 |
@lambday | trumped :D | 18:14 |
-!- besser82 [~besser82@fedora/besser82] has quit [Ping timeout: 268 seconds] | 18:14 | |
@HeikoS | haha man cracking here ;) | 18:16 |
@lambday | HeikoS: it was last night's beer talking maybe in the morning :D | 18:16 |
lisitsyn | lambday: ok I like the idea of no switch | 18:16 |
@lambday | lisitsyn: I hate that linalg::backend::fuckthisshit stuff | 18:16 |
@lambday | lisitsyn: check the API suggestions I proposed later as a comment | 18:17 |
lisitsyn | lambday: I have a suggestion for you how to develop the api | 18:17 |
@lambday | lisitsyn: sure man | 18:17 |
lisitsyn | it is - start porting some matrix intensive code | 18:17 |
lisitsyn | you will get all the troubles | 18:17 |
@lambday | lisitsyn: could you elaborate a bit? | 18:18 |
lisitsyn | lambday: just take any svm whatever | 18:18 |
lisitsyn | and try to port it to linalg | 18:18 |
lisitsyn | if it doesn't work - patch linalg :) | 18:18 |
@lambday | aha | 18:18 |
@lambday | yeah that sounds like a good plan | 18:18 |
@HeikoS | just got a good email for easy installation project | 18:18 |
@HeikoS | *finally* | 18:18 |
@HeikoS | really want this project to happen ;) | 18:19 |
lisitsyn | lambday: I mean it would be really hard to not miss some detail | 18:19 |
@HeikoS | lisitsyn: good point actually | 18:20 |
lisitsyn | lambday: https://github.com/shogun-toolbox/shogun/blob/develop/src/shogun/regression/LeastAngleRegression.cpp | 18:20 |
lisitsyn | try porting that :D | 18:20 |
@lambday | lisitsyn: problem with this approach is, we already need to have a good enough stuffs inside linalg to port existing things.. but that should be the goal | 18:20 |
@HeikoS | lambday: I thought this a while ago too ... rather than making thing broad, porting algorithm by algorithm might be a good idea | 18:20 |
lisitsyn | lambday: ah you can just skip what you've ported so far | 18:20 |
@HeikoS | lambday: so going deep first might be good | 18:20 |
lisitsyn | the thing is that mutates the framework | 18:20 |
lisitsyn | so it actually can do this thing | 18:21 |
@HeikoS | lisitsyn: I guess it needs to be both | 18:21 |
lisitsyn | otherwise we will get all these problems later | 18:21 |
@HeikoS | since high level design is also important | 18:21 |
@HeikoS | lisitsyn: ah man that LARS has cholesky updates implemented in it | 18:21 |
@HeikoS | ARGH | 18:21 |
lisitsyn | HeikoS: lol | 18:21 |
@HeikoS | this is needed in like 10 places in shogun | 18:22 |
lisitsyn | why not write good ol' cholesky | 18:22 |
@HeikoS | why is it in the f** lars class ;) | 18:22 |
lisitsyn | poor french | 18:22 |
@HeikoS | haha | 18:22 |
@lambday | holy shit man this is nightmare | 18:22 |
lisitsyn | haha | 18:22 |
lisitsyn | ok we need property based testing | 18:22 |
@lambday | I get scared every time I see cblas | 18:22 |
lisitsyn | otherwise you can't patch any function | 18:22 |
@HeikoS | lambday: lets do the draft for linalg, once the basic changes are in | 18:23 |
@lambday | good point | 18:23 |
@HeikoS | we should try to port whole algorithms | 18:23 |
@HeikoS | rather than adding single functions | 18:23 |
@HeikoS | but rather try to really make use of it | 18:23 |
@HeikoS | GP stuff is a good one for that | 18:23 |
@HeikoS | and things like LARS | 18:23 |
@lambday | HeikoS: you mean, the entire thing should reside inside linalg? | 18:24 |
@HeikoS | entire thing? | 18:24 |
@lambday | cholesky_delete, cholesky_insert | 18:24 |
@lambday | yeah these can be there | 18:24 |
@lambday | lisitsyn: dude you found some file to give an example | 18:25 |
@HeikoS | lambday: oh yes | 18:25 |
@HeikoS | lambday: cholesky updates are used everywhere | 18:25 |
lisitsyn | haha | 18:25 |
@HeikoS | my python code is full of them | 18:25 |
@HeikoS | haha | 18:25 |
@HeikoS | indeed | 18:25 |
@HeikoS | I gotta run now guys, dinner is ready | 18:25 |
sanuj | HeikoS: wait :P | 18:26 |
@HeikoS | sanuj: jo | 18:26 |
@lambday | HeikoS: I haven't yet cooked my dinner :( and it's 2300 in India | 18:26 |
@HeikoS | mine is warmed up ;) | 18:26 |
@HeikoS | from this morning | 18:26 |
@HeikoS | really gotta go | 18:26 |
@HeikoS | sanuj: gotta talk later | 18:26 |
sanuj | HeikoS: hi, i wanted to ask....shall i make separate pages for svm in cookbook? | 18:26 |
@HeikoS | Ill check more emails tonight | 18:26 |
sanuj | like linear, kernel etc | 18:26 |
@HeikoS | sanuj: yes | 18:26 |
sanuj | HeikoS: cool | 18:27 |
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has quit [Quit: Leaving.] | 18:27 | |
lisitsyn | lambday: have you looked at torch design? | 18:27 |
@lambday | lisitsyn: nope.. what's torch? | 18:27 |
lisitsyn | lambday: ah btw don't we need some tensorflow compatibility nowadays? | 18:27 |
sanuj | lambday: dp framework in lua | 18:27 |
lisitsyn | lambday: https://github.com/torch/torch7/blob/master/doc/tensor.md | 18:27 |
sanuj | it's good | 18:27 |
sanuj | dp - deep learning | 18:28 |
lisitsyn | how's that dp ;) | 18:28 |
lisitsyn | just kidding | 18:28 |
sanuj | lisitsyn: :P | 18:28 |
@lambday | yeah dp is misleading ;) | 18:28 |
@lambday | anyway what should I be looking for here? | 18:28 |
sanuj | lisitsyn: even i thought the same | 18:28 |
lisitsyn | lambday: well just general things maybe | 18:29 |
sanuj | lisitsyn: but then i used this for one of my projects | 18:29 |
sanuj | lisitsyn: https://github.com/nicholas-leonard/dp | 18:29 |
lisitsyn | I mean I think they have spent some time designing this | 18:29 |
lisitsyn | sanuj: that's funny everybody has a deep learning framework now | 18:30 |
lisitsyn | oh from wikipedia I just got to know yandex uses torch | 18:30 |
lisitsyn | :D | 18:30 |
sanuj | lisitsyn: yes, deep learning is very popular | 18:30 |
@lambday | it's funnier when they make the acronym out of (d)ee(p) as dp | 18:30 |
sanuj | lol | 18:30 |
lisitsyn | going to tell all my colleagues we're using torch :P | 18:31 |
@lambday | lisitsyn: hahaha | 18:31 |
sanuj | mlpack is also going to add deep learning algorithms | 18:31 |
lisitsyn | I think we should either drop our nns | 18:31 |
lisitsyn | or no idea | 18:32 |
lisitsyn | develop it | 18:32 |
@lambday | what's nns | 18:32 |
sanuj | neural nets | 18:32 |
sanuj | i guess | 18:32 |
lisitsyn | yeah | 18:32 |
@lambday | I am really bad at acronyms, aren't I | 18:32 |
lisitsyn | no | 18:32 |
sanuj | no :) | 18:33 |
sanuj | there are many libs for deep learning which are very good and popluar - caffe, torch, tensorflow, theano, keras, CNTK........ | 18:34 |
sanuj | lisitsyn: can't we use one of these in shogun? | 18:34 |
sanuj | like we use opencv | 18:34 |
sanuj | but i think that's a bad idea | 18:34 |
lisitsyn | sanuj: I think we should have fancy models not available in all these | 18:35 |
lisitsyn | like bayesian | 18:35 |
@lambday | sanuj: I don't think we're gonna use other libs for deep learning - we gotta develop one on our own | 18:35 |
lisitsyn | but not sure we should even have convolutional nn | 18:35 |
lisitsyn | because we will never be as good | 18:35 |
sanuj | cool | 18:36 |
sanuj | but if we are competitive then people will use it | 18:36 |
lisitsyn | but for bayesian stuff we should have all the basic stuff | 18:36 |
sanuj | i mean shogun users | 18:36 |
@lambday | lisitsyn: why won't we be ever as good? | 18:36 |
lisitsyn | lambday: they already put millions of hours there | 18:36 |
lisitsyn | I think a few people died there :D | 18:37 |
sanuj | yes and some of them support multiple gpu training | 18:37 |
@lambday | lisitsyn: okay so the problem is, we don't wanna die | 18:37 |
lisitsyn | people trained millions of models with caffe and stuff like that | 18:37 |
lisitsyn | we'd have to go this way again | 18:38 |
sanuj | and we have limited resources | 18:38 |
lisitsyn | doesn't sound possible for me | 18:38 |
sanuj | compared to google and facebook :P | 18:38 |
lisitsyn | we should take some short route | 18:38 |
@lambday | sanuj: you're forgetting baidu | 18:38 |
sanuj | lambday: oh yeah | 18:38 |
@lambday | lisitsyn: we don't wanna be excellent at a single thing, do we... | 18:39 |
lisitsyn | yeah but even to be competitive at neural networks we would need like 5 people working on that | 18:39 |
@lambday | amen to that | 18:40 |
@lambday | maybe next year | 18:40 |
lisitsyn | but I hope neural networks die next year | 18:40 |
lisitsyn | ;) | 18:40 |
@lambday | nah they are here to stay | 18:40 |
sanuj | agreed | 18:40 |
@lambday | I feel illiterate because I never worked on any | 18:40 |
lisitsyn | unfortunately hehe | 18:40 |
sanuj | everything is getting deep in ml :P | 18:40 |
@lambday | all this SVM and feature selection and other convex/submodular optimizatons - that's all I've worked on :( | 18:41 |
lisitsyn | I know all the stuff but I have no time | 18:41 |
sanuj | lambday: there are very cool deep learning courses online ;) | 18:41 |
sanuj | if you want to learn | 18:42 |
@lambday | gotta get into an ML company.. this oracle is sucking my blood out for nothing | 18:42 |
@lambday | sanuj: I hardly get any time bro :( | 18:42 |
lisitsyn | lambday: well I don't get gps as well as I do neural networks | 18:42 |
sanuj | lambday: even i'm graduating in June....got placed at samsung :( | 18:42 |
@lambday | after months I've been active doing some good work :'( | 18:42 |
lisitsyn | because you have to think a bit when it comes to gps | 18:43 |
lisitsyn | neural networks are dumb as hell | 18:43 |
@lambday | sanuj: congratulations | 18:43 |
@lambday | lisitsyn: really | 18:43 |
@lambday | that's another thing I don't know | 18:43 |
sanuj | lambday: not sure what to do :P | 18:43 |
sanuj | feel like doing research | 18:43 |
@lambday | got the book, got the lecture serieses, but hardly get time | 18:44 |
@lambday | sanuj: good idea.. I'm leaving my job | 18:44 |
sanuj | lambday: haha | 18:45 |
@lambday | sanuj: gotta get another one for now cause I'm getting married.. after that, I'm out! | 18:45 |
sanuj | lambday: oh good luck ;) | 18:45 |
sanuj | and congratulations :) | 18:45 |
lisitsyn | lambday: are you going to have that fancy thing with one thousand people? | 18:45 |
@lambday | sanuj: thanks :D | 18:45 |
sanuj | lisitsyn: haha | 18:46 |
@lambday | lisitsyn: pretty close - I am expecting 600+ people at lesat - like one of those bollywood movies | 18:46 |
lisitsyn | hahaha | 18:46 |
@lambday | withing singing and dancing and colors | 18:46 |
@lambday | haha yeah that's how we roll ;) | 18:46 |
@lambday | lisitsyn: I told HeikoS once about this 600+ people thing - he was shocked :D | 18:47 |
lisitsyn | hahah that's too funny | 18:47 |
sanuj | rofl | 18:47 |
lisitsyn | you don't even know them, do you? | 18:47 |
@lambday | lisitsyn: never talked to >= 85% of them in my life | 18:47 |
@lambday | and in case I did, I don't remember | 18:48 |
sanuj | lisitsyn: it's complicated | 18:48 |
sanuj | i'm also from india :) | 18:48 |
lisitsyn | sanuj: yeah I know :) | 18:48 |
lisitsyn | you're guys having fun there | 18:48 |
@lambday | lisitsyn: actually, over last few months, I started to kinda root for this large wedding thing - I guess that's the India in me talking :D | 18:49 |
lisitsyn | lambday: when is it happening? | 18:49 |
@lambday | lisitsyn: 22nd Jan 2017, if things go smooth | 18:50 |
lisitsyn | holy sheatz and you're already preparing? | 18:50 |
sanuj | lisitsyn: haha | 18:50 |
@lambday | lisitsyn: ...and we're late at that, as it turns out | 18:50 |
lisitsyn | well I guess that caused by that 600+ count | 18:50 |
@lambday | things get booked even before 1 year | 18:50 |
@lambday | and there are a lot of things to think of man, you won't understand | 18:51 |
@lambday | things **have** to be the way the bride wants it | 18:51 |
sanuj | lisitsyn: there will be many more events before lambday gets married | 18:51 |
lisitsyn | ohoha | 18:51 |
@lambday | sanuj: donald trump as US president | 18:51 |
lisitsyn | make shogun work again pleaze | 18:52 |
sanuj | lambday: are u planning to move to USA? | 18:52 |
sanuj | lisitsyn: +1 | 18:52 |
@lambday | for example | 18:52 |
@lambday | sanuj: it's tough, maybe I'll apply for PhD at some US schools but cannot settle there | 18:52 |
sanuj | lisitsyn: how difficult is that MCMC stan thing? assuming i don't know much about MCMC | 18:52 |
lisitsyn | sanuj: what thing exactly? | 18:53 |
sanuj | lambday: PhD is cool (y) | 18:53 |
lisitsyn | sanuj: the library itself? | 18:53 |
@lambday | sanuj: and tough! ;) | 18:53 |
sanuj | lisitsyn: https://github.com/shogun-toolbox/shogun/wiki/GSoC_2015_project_MCMC_Stan | 18:53 |
sanuj | this :) | 18:53 |
sanuj | i won't write a proposal for this | 18:54 |
lisitsyn | ah ok | 18:54 |
@lambday | sanuj: you're from which college? | 18:54 |
sanuj | but if i get selected for some other project, then i could learn MCMC also | 18:54 |
sanuj | lambday: IIT guwahati final year | 18:54 |
@lambday | sanuj: ah I've been there :D | 18:55 |
@lambday | nice place | 18:55 |
sanuj | lambday: i know you did your masters from IIT-B | 18:55 |
lisitsyn | sanuj: stan has a good codebase | 18:55 |
lisitsyn | I don't think its really tough | 18:55 |
sanuj | :) | 18:55 |
@lambday | sanuj: great girls! mostly bong professors | 18:55 |
sanuj | lisitsyn: okay | 18:55 |
sanuj | lambday: haha | 18:55 |
sanuj | lambday: i'm scared about graduating...dunno what i'll do | 18:56 |
lisitsyn | sanuj: all you need to know about MCMC is that it is what you use when you're lazy or can't compute some integral ;) | 18:56 |
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun | 18:56 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 18:56 | |
sanuj | lisitsyn: okay | 18:56 |
lisitsyn | sanuj: in this project the plan is to employ its solvers and autograd | 18:57 |
@lambday | sanuj: on don't worry you'll be fine.. figuring things out on the way - that's the key :) | 18:57 |
sanuj | lisitsyn: i was planning to submit my proposals for these 2 projects | 18:57 |
lisitsyn | HeikoS: you're back? | 18:57 |
sanuj | https://github.com/shogun-toolbox/shogun/wiki/GSoC_2016_project_approx_kernels | 18:57 |
sanuj | https://github.com/shogun-toolbox/shogun/wiki/GSoC_2015_plugin | 18:57 |
@HeikoS | lisitsyn: not really ;) but listening on one ear | 18:57 |
lisitsyn | sanuj: good! | 18:58 |
sanuj | lambday: thanks | 18:58 |
lisitsyn | sanuj: plugin thing would give one all the fame | 18:58 |
sanuj | lisitsyn: yeah, this is my first priority...guess this is important for shogun | 18:59 |
sanuj | lisitsyn: and i would get to learn about ml in general | 18:59 |
@lambday | go plugin | 19:00 |
lisitsyn | HeikoS: are our gp classification sampling based? | 19:01 |
-!- Nebula_ [~chatzilla@1.39.61.226] has joined #shogun | 19:01 | |
@HeikoS | lisitsyn: no | 19:01 |
lisitsyn | HeikoS: what kind of approximation? | 19:01 |
@HeikoS | ep and mostly variational | 19:01 |
lisitsyn | I see | 19:01 |
lisitsyn | HeikoS: btw does sampling works at all? | 19:02 |
lisitsyn | HeikoS: takes unreasonable time? | 19:02 |
lisitsyn | or? %) | 19:02 |
lisitsyn | :) | 19:02 |
@HeikoS | lisitsyn: it works very nicely | 19:03 |
@HeikoS | most general method | 19:03 |
@HeikoS | bust obviosly slower | 19:03 |
lisitsyn | HeikoS: I see | 19:03 |
@HeikoS | variational methods are quick but not accurate | 19:03 |
@HeikoS | if you need accuracy of uncertainty, mcmc is king | 19:03 |
@HeikoS | lisitsyn: but if you only care about predictive RMSE, then EP or variational is totally fine | 19:04 |
@HeikoS | if model is complex, variational methods fail | 19:04 |
@HeikoS | but for GPS ,which are simple, all good | 19:04 |
@HeikoS | really depends | 19:04 |
sanuj | HeikoS: i got this education from a shogun talk on youtube by emtiyaz :) | 19:04 |
@HeikoS | these methods are not better than one another | 19:04 |
@HeikoS | sanuj: haha :) | 19:04 |
sanuj | i watched it today | 19:04 |
@HeikoS | nice | 19:04 |
@HeikoS | emt was at UCL and gave a talk on Friday | 19:04 |
lisitsyn | HeikoS: nice to know, thanks! | 19:05 |
sanuj | cool | 19:05 |
lisitsyn | HeikoS: I think we can do some kind of experiment | 19:05 |
@HeikoS | lisitsyn: which is? | 19:05 |
lisitsyn | I can implement a few methods to use our class_list.cpp | 19:05 |
lisitsyn | and then disable everything in swig, but these methods | 19:05 |
lisitsyn | this is not plugins but we get this narrow interface we expect from plugins | 19:06 |
lisitsyn | we're already collecting classes by their names | 19:06 |
@HeikoS | whats the goal? | 19:06 |
@HeikoS | ah I see | 19:06 |
@HeikoS | to design the interface? | 19:06 |
-!- Nebula_ [~chatzilla@1.39.61.226] has left #shogun [] | 19:06 | |
@HeikoS | good entrance task! | 19:06 |
lisitsyn | shogun.classifier("LibLinear") | 19:06 |
@HeikoS | lisitsyn: yeah | 19:07 |
sanuj | nice talk guys, i gotta leave. | 19:07 |
lisitsyn | sanuj: see ya | 19:07 |
@HeikoS | sanuj: take care! | 19:07 |
sanuj | bye :) | 19:07 |
@HeikoS | lisitsyn: I also goota go soon | 19:07 |
lisitsyn | HeikoS: I think I have to try it first because we can't merge it yet | 19:07 |
sanuj | some PRs coming tomorrow | 19:07 |
@HeikoS | lisitsyn: yeah definitely | 19:07 |
lisitsyn | so not really a good task for a student | 19:07 |
@HeikoS | it is a good start for a project | 19:07 |
-!- mizari [~mizari@95-174-213-100.nts.su] has quit [Read error: Connection reset by peer] | 19:07 | |
@HeikoS | lisitsyn: well | 19:08 |
-!- sanuj [0e8bc402@gateway/web/freenode/ip.14.139.196.2] has quit [Quit: Page closed] | 19:08 | |
lisitsyn | they need merged code ;) | 19:08 |
@HeikoS | after your draft is done | 19:08 |
lisitsyn | ok lets see | 19:08 |
@HeikoS | lisitsyn: feature branch drafts are also fine I guess? | 19:08 |
lisitsyn | yeap | 19:08 |
lisitsyn | that's what I am going to try | 19:08 |
@lambday | very useful | 19:08 |
@HeikoS | yeah thats the way to go I guess | 19:09 |
@HeikoS | worked well for the cookbook as well | 19:09 |
@HeikoS | the build was only broken a bit | 19:09 |
@HeikoS | lisitsyn: speaking of, can you fix ctags on bsd? | 19:09 |
@HeikoS | ok gotta really go now .... see you later | 19:09 |
lisitsyn | HeikoS: ctags - ok, will check | 19:10 |
lisitsyn | see you | 19:10 |
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has quit [Quit: Leaving.] | 19:11 | |
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun | 19:13 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 19:13 | |
-!- lambday [6a33f351@gateway/web/freenode/ip.106.51.243.81] has quit [Ping timeout: 252 seconds] | 19:17 | |
-!- lambday [6a33f351@gateway/web/freenode/ip.106.51.243.81] has joined #shogun | 19:34 | |
-!- mode/#shogun [+o lambday] by ChanServ | 19:34 | |
-!- curiousguy13 [~curiousgu@120.57.233.223] has quit [Remote host closed the connection] | 19:34 | |
@lambday | lisitsyn: shit shogun doesn't build on my laptop! | 19:35 |
-!- Saurabh7 [~Saurabh7@103.225.100.51] has joined #shogun | 19:58 | |
-!- witness_ [uid10044@gateway/web/irccloud.com/x-lxxoegdxpfzkglbv] has quit [Quit: Connection closed for inactivity] | 20:08 | |
-!- Saurabh7 [~Saurabh7@103.225.100.51] has quit [Ping timeout: 248 seconds] | 20:31 | |
-!- livus [~suvil@196.207.84.20] has quit [Ping timeout: 260 seconds] | 20:32 | |
lisitsyn | lambday: why? | 20:46 |
@lambday | lisitsyn: some issue with gpumatrix.. GPUMatrix.cpp:40: error: redeclared with 4 template parameters | 20:49 |
@lambday | I have outdated viennacl | 20:49 |
-!- GandalfTheWizard [~ian@112.10.107.176] has quit [Ping timeout: 244 seconds] | 21:00 | |
-!- GandalfTheWizard [~ian@112.10.107.176] has joined #shogun | 21:02 | |
-!- GandalfTheWizard [~ian@112.10.107.176] has quit [Ping timeout: 248 seconds] | 21:09 | |
-!- sonne|osx [~sonne@x4db4d6d7.dyn.telefonica.de] has joined #shogun | 21:11 | |
@lambday | lisitsyn: can't we have sub namespaces under shogun? seems like class-list.cpp only works with shogun namespace | 21:21 |
lisitsyn | lambday: well we'd have to patch it then I guess | 21:22 |
@lambday | lisitsyn: seems so.. I just noticed this | 21:23 |
@lambday | lisitsyn: how bad it would be if I add some c++14 checks in cmake and enable -std=c++14 if a compiler supports this? not saying that we'd make it hardcode requirement | 21:25 |
lisitsyn | lambday: no idea | 21:26 |
lisitsyn | we are not C++11 still.. | 21:26 |
@lambday | lisitsyn: would it hurt? it's not like build will break if it's not there | 21:27 |
lisitsyn | lambday: I don't think it would hurt | 21:27 |
@lambday | lisitsyn: actually the auto param type in lambdas is quite useful.. | 21:28 |
@lambday | probably that's the only c++14 stuff I used in my code | 21:28 |
@lambday | nah it's better if I change those | 21:31 |
-!- Nebula_ [~chatzilla@1.39.61.32] has joined #shogun | 21:32 | |
-!- Nebula_ [~chatzilla@1.39.61.32] has quit [Client Quit] | 21:34 | |
-!- sonne|osx [~sonne@x4db4d6d7.dyn.telefonica.de] has quit [Quit: sonne|osx] | 21:41 | |
lisitsyn | lambday: still here? | 21:54 |
@lambday | lisitsyn: yessir | 21:54 |
lisitsyn | lambday: going to show you something | 21:54 |
@lambday | lisitsyn: sure what is it | 21:54 |
-!- GandalfTheWizard [~ian@112.10.107.176] has joined #shogun | 21:55 | |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 21:56 | |
shogun-notifier- | shogun: Sergey Lisitsyn :narrow_interface * a9d5428 / / (3 files): https://github.com/shogun-toolbox/shogun/commit/a9d5428912bc08f11d0937b18dccfc5bc1127c50 | 21:56 |
shogun-notifier- | shogun: Introduce a function to create SGOBject by name | 21:56 |
lisitsyn | lambday: https://github.com/shogun-toolbox/shogun/commit/a9d5428912bc08f11d0937b18dccfc5bc1127c50 | 21:56 |
lisitsyn | ah shogun-notifier- made this for me | 21:56 |
lisitsyn | lambday: check the unit-test | 21:56 |
@lambday | lisitsyn: hehe classy.. this is useful! | 21:57 |
shogun-buildbot | build #3668 of deb1 - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun/builds/3668 blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com> | 21:57 |
@lambday | lisitsyn: maybe useful for class_list | 21:57 |
lisitsyn | lambday: well it uses the class_list | 21:58 |
lisitsyn | lambday: the thing is | 21:58 |
lisitsyn | I want to try to add some more methods and then disable *everything* in swig :) | 21:58 |
lisitsyn | and see how fast it is | 21:58 |
@lambday | lisitsyn: how do you plan to use this one for that? | 21:59 |
lisitsyn | lambday: we could change how objects are created under the hood later | 22:00 |
lisitsyn | so it uses plugins | 22:00 |
lisitsyn | HeikoS: oh u here? | 22:01 |
lisitsyn | lambday: there is a thing with new_sgserializable | 22:01 |
lisitsyn | there is a hack for templated classes | 22:02 |
@lambday | lisitsyn: that's what I was thinking - how to instantiate template classes | 22:02 |
@lambday | what's the hack | 22:02 |
lisitsyn | lambday: some enum | 22:02 |
lisitsyn | lambday: check the definition at shogun/base/class_list.h | 22:03 |
@lambday | lisitsyn: only for primitive typees | 22:03 |
lisitsyn | lambday: yes | 22:03 |
lisitsyn | I don't think we need templates for non-primitives | 22:03 |
@lambday | lisitsyn: in the present stuff I am working on, I used a few :| | 22:05 |
lisitsyn | lambday: should they be interfaced? | 22:05 |
@lambday | lisitsyn: they are base classes, so, yes | 22:06 |
lisitsyn | meh | 22:06 |
@lambday | lisitsyn: I didn't put all these stuffs under shogun yet... so I missed to consider a few things | 22:07 |
@lambday | now having the pain | 22:07 |
lisitsyn | lambday: I see | 22:07 |
@lambday | lisitsyn: e.g. https://github.com/lambday/flash/blob/develop/src/flash/statistics/HypothesisTest.h#L42 | 22:08 |
lisitsyn | lambday: what's test type? | 22:10 |
@lambday | lisitsyn: custom struct | 22:10 |
lisitsyn | lambday: can you make it dynamic? ;) | 22:10 |
@lambday | https://github.com/lambday/flash/blob/develop/src/flash/statistics/internals/TestTypes.h | 22:10 |
@lambday | lisitsyn: doable - painful :D | 22:11 |
lisitsyn | lambday: I mean you don't really get much speedup by that | 22:12 |
lisitsyn | but interfacing is painful | 22:12 |
@lambday | lisitsyn: yeah that's what I am noticing | 22:12 |
@lambday | lisitsyn: BTW when the template param is another SGObject, we're good I guess | 22:12 |
@lambday | PT_SGOBJECT is there | 22:12 |
lisitsyn | lambday: yeah somewhat ok | 22:13 |
@lambday | not saying that I'll make these test types SGObjects.. but it's about another case where I used a static polymorphism | 22:13 |
lisitsyn | lambday: I'd keep static polymorphism for something internal ;) | 22:14 |
lisitsyn | lambday: oh I just realised 'object' is not a good name hah | 22:16 |
lisitsyn | in the python context | 22:16 |
@lambday | lisitsyn: how about instance? | 22:20 |
lisitsyn | lambday: yeah good | 22:21 |
@lambday | it works like a factory - instance sounds reasonable | 22:21 |
lisitsyn | lambday: anyway, this function is not expected to be used directly | 22:21 |
lisitsyn | I will add some classifier which is just instance + cast | 22:21 |
lisitsyn | and stuff like that | 22:21 |
-!- besser82 [~besser82@fedora/besser82] has joined #shogun | 22:39 | |
-!- mode/#shogun [+o besser82] by ChanServ | 22:39 | |
-!- besser82 [~besser82@fedora/besser82] has quit [Ping timeout: 268 seconds] | 23:03 | |
@HeikoS | lisitsyn: im here now | 23:11 |
@HeikoS | lambday: hi | 23:11 |
@HeikoS | had a discussion? | 23:11 |
lisitsyn | HeikoS: I am currently experimenting with disabling everything but this very method | 23:11 |
lisitsyn | object(const char*) | 23:11 |
@lambday | HeikoS: hey.. yeah | 23:11 |
@HeikoS | uh | 23:11 |
@HeikoS | scary stuff ;) | 23:11 |
lisitsyn | HeikoS: without actual classes swig compiles instantly :) | 23:13 |
lisitsyn | that's predictable tho | 23:13 |
@HeikoS | lisitsyn: thats also desirable ;) | 23:17 |
@HeikoS | lisitsyn: that will be a good draft for gsoc project | 23:17 |
@HeikoS | lisitsyn: is it that tricky? | 23:18 |
lisitsyn | HeikoS: it seems not really | 23:20 |
lisitsyn | HeikoS: I am now having fun with swig segfault | 23:21 |
lisitsyn | usual mac problem now | 23:21 |
@HeikoS | haha | 23:21 |
lisitsyn | I don't even feel angry anymore | 23:22 |
@lambday | HeikoS: merging this thing to shogun is gonna give me some pain | 23:24 |
@HeikoS | lambday: the flash? | 23:24 |
@lambday | HeikoS: yep | 23:24 |
@HeikoS | lambday: whats the painkiller? :D | 23:24 |
@lambday | HeikoS: apparently, patience :| | 23:24 |
@lambday | sent a patch though - with the internals.. | 23:25 |
@lambday | will send SGObjects patches one by one | 23:25 |
* lisitsyn compiles python because it is easier than to use system one | 23:26 | |
@HeikoS | lambday: haha | 23:28 |
@HeikoS | lambday: in a feature branch? | 23:29 |
@lambday | HeikoS: yes | 23:29 |
@HeikoS | lambday: cool, travis passes? | 23:29 |
@lambday | HeikoS: https://github.com/shogun-toolbox/shogun/pull/3103 | 23:29 |
@lambday | it was doing some previous build, last time I checked | 23:29 |
@HeikoS | lambday: hey btw just push into this, no nee to do PRs in this feature branch | 23:29 |
@lambday | I am adding documentation and making the code look Shogun style | 23:29 |
@lambday | HeikoS: oh | 23:30 |
@HeikoS | merge it, then I can compile here :) | 23:30 |
@lambday | HeikoS: cool | 23:30 |
shogun-notifier- | shogun: lambday :feature/bigtest * ba67219 / src/shogun/statistics/experimental/internals/ (34 files): https://github.com/shogun-toolbox/shogun/commit/ba67219ea44488a525b46c48db2099b4b8345409 | 23:30 |
shogun-notifier- | shogun: added first version of bigtesting framework | 23:30 |
shogun-notifier- | shogun: Soumyajit De :feature/bigtest * fbe78df / src/shogun/statistics/experimental/internals/ (34 files): https://github.com/shogun-toolbox/shogun/commit/fbe78df6ceb7c41b6ca0d9b2c6849e9e4ee69157 | 23:31 |
shogun-notifier- | shogun: Merge pull request #3103 from lambday/feature/bigtest | 23:31 |
shogun-notifier- | shogun: | 23:31 |
shogun-notifier- | shogun: added first version of bigtesting framework | 23:31 |
@lambday | HeikoS: didn't add any of the SGObjects yet.. so not yet usable :( | 23:31 |
@HeikoS | lambday: ok | 23:31 |
@HeikoS | ill compile anyways | 23:31 |
@lambday | got gazillions of errors | 23:31 |
shogun-buildbot | build #3669 of deb1 - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun/builds/3669 blamelist: lambday <heavensdevil6909@gmail.com>, Soumyajit De <heavensdevil6909@gmail.com> | 23:32 |
@HeikoS | compiling ... | 23:34 |
lisitsyn | HeikoS: we need features 2.0 | 23:38 |
lisitsyn | separating features layout and structure | 23:38 |
lisitsyn | to mix up categorical and numeric things | 23:38 |
@HeikoS | lisitsyn: yes please | 23:40 |
@HeikoS | lisitsyn: maybe can be optional part of plugin framework | 23:40 |
@HeikoS | but not sure | 23:40 |
@HeikoS | maybe something for later | 23:40 |
@HeikoS | lambday: compiles | 23:40 |
@HeikoS | but I cant run I guess? | 23:40 |
@lambday | HeikoS: nah the wrappers are not there yet | 23:41 |
@lambday | HeikoS: BTW can you please try compiling flash once? it's the same API.. | 23:42 |
@lambday | just wanna make sure that the sanity tests work | 23:42 |
@lambday | HeikoS: just run the sanity script | 23:42 |
@HeikoS | lambday: ok cloning | 23:42 |
shogun-notifier- | shogun: youssef :develop * 18686e5 / src/shogun/multiclass/QDA.cpp: https://github.com/shogun-toolbox/shogun/commit/18686e5cc7850c8c2523ab352f8fa52a79e722ae | 23:45 |
shogun-notifier- | shogun: remove rotations transpose | 23:45 |
shogun-notifier- | shogun: Heiko Strathmann :develop * a77c561 / src/shogun/multiclass/QDA.cpp: https://github.com/shogun-toolbox/shogun/commit/a77c56150af82255f3c163df267effe4af89adc6 | 23:45 |
shogun-notifier- | shogun: Merge pull request #3102 from youssef-emad/develop | 23:45 |
shogun-notifier- | shogun: | 23:45 |
shogun-notifier- | shogun: Remove transpose of rotations in QDA | 23:45 |
lisitsyn | SHEATZ swig swig swig | 23:45 |
lisitsyn | Y U NO WORK | 23:45 |
@HeikoS | lambday: I get lots of no such file or dir | 23:46 |
@lambday | lisitsyn: changing that base template to non-template.. making the enums as static int.. essentially serves the same purpose | 23:46 |
@lambday | HeikoS: err.. this is an old school makefile.. maybe I missed something | 23:46 |
shogun-buildbot | build #3670 of deb1 - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun/builds/3670 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, youssef <youssef.emad.attia@gmail.com> | 23:47 |
@lambday | HeikoS: which dir is missing | 23:47 |
@HeikoS | https://gist.github.com/karlnapf/10f5d794b53451e58abc | 23:48 |
@HeikoS | https://gist.github.com/karlnapf/10f5d794b53451e58abc | 23:48 |
@HeikoS | https://gist.github.com/karlnapf/10f5d794b53451e58abc | 23:48 |
@HeikoS | ah sorry | 23:48 |
@HeikoS | lag | 23:48 |
@lambday | HeikoS: oh crap it won't compile without c++14.. your compiler doesn't have that! :| | 23:48 |
@HeikoS | haha | 23:48 |
@lambday | HeikoS: I'll finish the merging with Shogun tomorrow.. | 23:48 |
@HeikoS | i can install one | 23:48 |
@lambday | then we can check | 23:49 |
@HeikoS | got c++ 4.8.4-2 | 23:49 |
lisitsyn | I want to see viktor's face when you merge C++14 code | 23:49 |
lisitsyn | HeikoS: your grandma C++ | 23:49 |
@lambday | lisitsyn: I swear I don't | 23:49 |
@lambday | I am using 5.3 | 23:49 |
@HeikoS | lisitsyn: hey btw we should push the c++11 thing soon | 23:49 |
lisitsyn | how? | 23:49 |
@HeikoS | sonney2k: is also for it | 23:49 |
@HeikoS | I think wait until next year and then do it | 23:49 |
@HeikoS | lets face it, nobody uses shogun anyways | 23:50 |
@HeikoS | so no need to block ourselves | 23:50 |
lisitsyn | yes | 23:50 |
@HeikoS | I mean I am all in for multiplatform and compatible etc etc | 23:50 |
@HeikoS | but with this, I think we can compromise | 23:50 |
lisitsyn | yes if you're grep | 23:50 |
lisitsyn | or curl | 23:50 |
@HeikoS | got more important issues | 23:50 |
lisitsyn | :D | 23:50 |
@HeikoS | yeah exactly | 23:50 |
@HeikoS | but we can do it after gsoc | 23:51 |
@lambday | would be great to get rid of those macros | 23:51 |
@HeikoS | one thing for sure: I dont want these double implementations | 23:51 |
lisitsyn | I give up this shit doesn't work again | 23:51 |
@HeikoS | haha | 23:51 |
lisitsyn | swig + mac = segfault | 23:51 |
@HeikoS | lambday: just reading your thing in detail and laughing here | 23:53 |
@lambday | HeikoS: which one? | 23:53 |
@lambday | ah linalg :D | 23:53 |
@HeikoS | lambday: yeah | 23:54 |
@HeikoS | pretty_please_put_data_on_gpu_if_possible_and_yes_I_know_it_is_costly | 23:54 |
@HeikoS | thats a good method name | 23:54 |
@HeikoS | lisitsyn would like it | 23:54 |
@lambday | hahahahaha | 23:54 |
@lambday | got the inspiration from eigen guys | 23:54 |
lisitsyn | haha | 23:54 |
@lambday | #define YES_I_KNOW_SPARSE_MODULE_IS_NOT_STABLE_YET | 23:54 |
-!- travis-ci [~travis-ci@ec2-54-144-82-39.compute-1.amazonaws.com] has joined #shogun | 23:55 | |
travis-ci | it's Wu Lin's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/117327021 | 23:55 |
-!- travis-ci [~travis-ci@ec2-54-144-82-39.compute-1.amazonaws.com] has left #shogun [] | 23:55 | |
@HeikoS | yes | 23:55 |
@HeikoS | lambday: hey I agree to change this soon, before we get too many patches and students get used to linalg | 23:55 |
@HeikoS | already happening, we have like 3-5 patches on this | 23:55 |
@lambday | HeikoS: yeah.. this is the first thing I am gonna do after I finish merging flash | 23:56 |
@HeikoS | great | 23:56 |
@HeikoS | lambday: ah man I love these second details you write there | 23:56 |
@HeikoS | linalg::elementwise(m).custom([](auto& v) { v = something_fun(v); }); | 23:56 |
@lambday | HeikoS: BTW how did you like the proposed API for elementwise | 23:56 |
@HeikoS | very good | 23:56 |
@lambday | HeikoS: yeah these would be neat | 23:56 |
@HeikoS | haha thats what I just meant | 23:56 |
@lambday | will create issues for students to achieve these effects | 23:57 |
@HeikoS | lambday: yeah thats a good idea | 23:57 |
@HeikoS | lambday: maybe we can do this in another feature branch and ppl can send their patches there? | 23:57 |
@HeikoS | lambday: not sure how to best organise this | 23:57 |
@HeikoS | lambday: should happen before gsoc start definitely | 23:58 |
@lambday | HeikoS: absolutely.. I won't take more than a week to do the rafactoring.. I'll be merging in some feature branch.. but then I think it's better to put it back to dev | 23:58 |
@lambday | (similar to your argument for shogun) no one uses linalg anyways.. | 23:59 |
@HeikoS | lambday: yeah haha | 23:59 |
@lambday | so we won't break anything hopefully | 23:59 |
@HeikoS | lambday: should be ok | 23:59 |
@HeikoS | we have the tests etc | 23:59 |
@HeikoS | linalg has like 100% coverage | 23:59 |
@lambday | yeah | 23:59 |
@lambday | HeikoS: getting rid of the NATIVEBLABLA thing would be the biggest gain API-wise IMO | 23:59 |
--- Log closed Mon Mar 21 00:00:05 2016 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!