IRC logs of #shogun for Monday, 2018-07-09

--- Log opened Mon Jul 09 00:00:00 2018
-!- HeikoS [~heiko@2a00:23c5:e10a:5c00:e59f:8ac6:d583:5195] has quit [Quit: Leaving.]00:48
-!- iglesiasg [6cab80bc@gateway/web/freenode/ip.108.171.128.188] has joined #shogun09:17
-!- wuwei [wuweilinma@gateway/shell/matrix.org/x-hlmammtjwckyuraq] has quit [Ping timeout: 240 seconds]10:04
-!- wuwei [wuweilinma@gateway/shell/matrix.org/x-vorcchhjubiijvnq] has joined #shogun10:32
-!- HeikoS [~heiko@2a00:23c5:e10a:5c00:e59f:8ac6:d583:5195] has joined #shogun10:51
-!- mode/#shogun [+o HeikoS] by ChanServ10:51
iglesiasgHeikoS: https://blog.jupyter.org/interactive-workflows-for-c-with-jupyter-fe9b54227d9210:57
iglesiasg^ check out the gifs, very nice10:58
lisitsyniglesiasg: wow10:58
lisitsynHeikoS: https://github.com/shogun-toolbox/shogun/pull/4364 coreml!10:58
lisitsyniglesiasg: ^10:58
lisitsyn:)10:58
iglesiasgI used cling last year in the command line10:58
iglesiasgnice also that is with jupyter10:58
iglesiasgolala that pr \o/10:59
lisitsyniglesiasg: have you seen we've got lazy parameters?11:09
lisitsynbasically that's all we need for plugins11:10
iglesiasgI saw the pr yeah11:23
iglesiasgbut I haven't really delved deep into it11:23
iglesiasgso what is it about, lazy parameters?11:23
iglesiasgfrom lazy, I thought it would be something like lazy evaluation in Eigen11:24
-!- HeikoS [~heiko@2a00:23c5:e10a:5c00:e59f:8ac6:d583:5195] has quit [Quit: Leaving.]11:54
iglesiasglisitsyn: ^13:06
lisitsyniglesiasg: uhmm I think I missed some message13:07
lisitsyniglesiasg: the last one I see is about Eigen13:07
iglesiasgyeah13:07
iglesiasgoh I repeat13:07
lisitsynis it?13:07
-!- HeikoS [~heiko@untrust-out.swc.ucl.ac.uk] has joined #shogun13:07
-!- mode/#shogun [+o HeikoS] by ChanServ13:07
lisitsynah13:07
lisitsynok13:07
lisitsynok so lazy is for cases13:07
iglesiasgso what is the main idea of the lazy parameters?13:07
lisitsynwhen your machine can compute something useful13:07
lisitsynbut you don't want to compute it right now13:07
lisitsynit might be some oligomer matrix of some crazy svm13:08
lisitsynthat is expensive to compute13:08
iglesiasgI see I see13:08
lisitsynI say it is useful for plugins because13:08
lisitsynnow we can add something that is not in the base class13:08
lisitsynso it is a replacement for a method with no parameters13:08
lisitsynHeikoS: coreml!13:08
@HeikoSlisitsyn: jojo13:09
@HeikoSso what is it exactly what you did there?13:09
lisitsynHeikoS: they are compiled with protoc13:09
@HeikoSlisitsyn:  I think viktor was one writing an execution library for coreml13:09
lisitsynyes but that's pure cmake thing13:09
lisitsynjust first step to integrate that13:09
@HeikoSok cool13:09
lisitsynI don't know how it would look like in our code yet13:10
lisitsynprobably it would be some convert(CMachine)13:10
lisitsynHeikoS: I am not sure coreml should stay in interface though13:11
@HeikoSyou mean to export as coreml?13:11
lisitsynyes13:11
@HeikoSor to run coreml stuff using shogun?13:11
lisitsynI'd focus on export13:12
@HeikoSI also like the idea of importing a coreml model from keras and then use it in shogun13:12
lisitsynHeikoS: running is possible with tf lite and coreml in ios13:12
@HeikoSbut yeah13:12
@HeikoSexport is more important13:12
@HeikoSI think the interface thing could be sweet13:12
@HeikoSrun a coreml model in java13:12
@HeikoSsee what I mean?13:12
lisitsynyeah but a lot of details to be solved13:13
lisitsyne.g.13:13
lisitsyncoreml is in interface now13:13
lisitsynbut do we export it using swig?13:13
lisitsynif so, do we generate protobufs for all the target languages?13:13
lisitsyn:)13:13
@HeikoSwell13:13
@HeikoSi think this should be a difference code base than shogun13:14
@HeikoSjust re-use the swig stuff13:14
@HeikoSbut yeah lots of details13:14
@HeikoSI think viktor prototyped something for that already13:14
@HeikoSbased on arraw13:14
@HeikoSarrow13:14
lisitsynok then lets wait13:14
@HeikoSbut yeah i like export definitelty13:14
lisitsynis he on some vacation?13:14
-!- travis-ci [~travis-ci@ec2-54-92-247-212.compute-1.amazonaws.com] has joined #shogun15:39
travis-ciit's Sergey Lisitsyn's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/40174536615:39
-!- travis-ci [~travis-ci@ec2-54-92-247-212.compute-1.amazonaws.com] has left #shogun []15:39
-!- travis-ci [~travis-ci@ec2-54-145-84-116.compute-1.amazonaws.com] has joined #shogun15:52
travis-ciit's Sergey Lisitsyn's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/40174536615:52
-!- travis-ci [~travis-ci@ec2-54-145-84-116.compute-1.amazonaws.com] has left #shogun []15:52
-!- iglesiasg [6cab80bc@gateway/web/freenode/ip.108.171.128.188] has quit [Quit: Page closed]17:22
-!- iglesias [6cab80bd@gateway/web/freenode/ip.108.171.128.189] has joined #shogun18:01
-!- iglesias [6cab80bd@gateway/web/freenode/ip.108.171.128.189] has quit [Quit: Page closed]18:09
-!- HeikoS [~heiko@untrust-out.swc.ucl.ac.uk] has quit [Ping timeout: 245 seconds]19:18
-!- HeikoS [~heiko@host86-172-81-177.range86-172.btcentralplus.com] has joined #shogun20:12
-!- mode/#shogun [+o HeikoS] by ChanServ20:12
-!- HeikoS [~heiko@host86-172-81-177.range86-172.btcentralplus.com] has quit [Quit: Leaving.]20:31
-!- HeikoS [~heiko@host86-172-81-177.range86-172.btcentralplus.com] has joined #shogun20:40
-!- mode/#shogun [+o HeikoS] by ChanServ20:40
-!- witness_ [uid10044@gateway/web/irccloud.com/x-hjtdyzklusfjcbps] has quit [Quit: Connection closed for inactivity]20:46
--- Log closed Tue Jul 10 00:00:02 2018

Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!