--- Log opened Wed Mar 06 00:00:38 2019 | ||
-!- wiking [~wiking@huwico/staff/wiking] has quit [Remote host closed the connection] | 01:13 | |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 01:33 | |
-!- mode/#shogun [+o wiking] by ChanServ | 01:33 | |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Remote host closed the connection] | 02:17 | |
-!- callmetesla [31ccd6b6@gateway/web/freenode/ip.49.204.214.182] has joined #shogun | 02:31 | |
-!- callmetesla [31ccd6b6@gateway/web/freenode/ip.49.204.214.182] has quit [Client Quit] | 02:32 | |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 04:18 | |
-!- mode/#shogun [+o wiking] by ChanServ | 04:18 | |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Ping timeout: 252 seconds] | 04:22 | |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 04:39 | |
-!- mode/#shogun [+o wiking] by ChanServ | 04:39 | |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Remote host closed the connection] | 04:54 | |
-!- taichi303j [672c33c8@gateway/web/freenode/ip.103.44.51.200] has joined #shogun | 04:57 | |
-!- taichi303j [672c33c8@gateway/web/freenode/ip.103.44.51.200] has quit [Client Quit] | 04:59 | |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 05:37 | |
-!- mode/#shogun [+o wiking] by ChanServ | 05:37 | |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Remote host closed the connection] | 06:20 | |
-!- ashwath [0e8be4dc@gateway/web/freenode/ip.14.139.228.220] has joined #shogun | 06:27 | |
ashwath | Hello | 06:27 |
---|---|---|
-!- ashwath [0e8be4dc@gateway/web/freenode/ip.14.139.228.220] has quit [Ping timeout: 256 seconds] | 06:34 | |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 06:35 | |
-!- mode/#shogun [+o wiking] by ChanServ | 06:35 | |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Ping timeout: 255 seconds] | 06:39 | |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 06:53 | |
-!- mode/#shogun [+o wiking] by ChanServ | 06:53 | |
-!- ashwath [0e8be4dc@gateway/web/freenode/ip.14.139.228.220] has joined #shogun | 08:26 | |
-!- gf712 [c13cdcfd@gateway/web/freenode/ip.193.60.220.253] has joined #shogun | 08:41 | |
gf712 | wiking: hi | 08:46 |
@wiking | hi | 08:47 |
gf712 | I was thinking that if we wanted we could make the factories more pythonic | 08:48 |
gf712 | rather than passing strings | 08:48 |
gf712 | we can just add them as class to the module at runtime | 08:49 |
gf712 | so would go back to sg.MyAlgo, but still using factories at it's core | 08:50 |
@wiking | the q here is | 09:22 |
@wiking | whether this could be done across all the interfaces | 09:22 |
@wiking | or this becomes haxors for each interface | 09:22 |
gf712 | hmmm in theory all interfaces could support this I think | 09:24 |
gf712 | I mean all it would do would be add stuff to the module | 09:24 |
gf712 | which is just a hash table in the stack, which all interpreted languages use right? | 09:25 |
@wiking | ye | 09:32 |
@wiking | but i mean | 09:32 |
@wiking | the problem is that this you cannot do with java | 09:32 |
@wiking | or any jvm | 09:32 |
-!- geektoni [c1cdd24a@gateway/web/freenode/ip.193.205.210.74] has joined #shogun | 09:32 | |
@wiking | (monkeypatching the module) | 09:32 |
@wiking | not sure about ruby | 09:32 |
@wiking | js for sure | 09:32 |
@wiking | because this is what you suggest | 09:32 |
@wiking | have the module | 09:33 |
@wiking | and then monkeypatch it | 09:33 |
@wiking | but then again | 09:33 |
@wiking | how do you wanna do the runtime extension of sg module | 09:33 |
@wiking | when the dll is not yet loaded into the memory | 09:33 |
@wiking | say we have the plugins framework | 09:33 |
@wiking | and you do | 09:33 |
@wiking | import shogun | 09:33 |
@wiking | this loads in the bare shogun-core with only interfaces etc | 09:33 |
@wiking | the ideas that when you do | 09:33 |
@wiking | shogun.machine("asdf") | 09:34 |
@wiking | then there's a dynamic loading of a libasdf.so in the backkend | 09:34 |
@wiking | that loads the shared lib that contains asdf machine | 09:34 |
gf712 | well you can do it lazily right? | 09:34 |
@wiking | and then just passes back the new instance of asdf | 09:34 |
@wiking | you say that instead of this | 09:34 |
gf712 | the things you see in the module is just a partial function | 09:35 |
@wiking | import shogun.asdf | 09:35 |
@wiking | and then where and who does the things mentioned above? | 09:35 |
@wiking | if not in dictionary: try load? :) | 09:35 |
gf712 | ok, I need to understand this better | 09:36 |
gf712 | when you load the .so what would you know in the plugin approach? | 09:36 |
gf712 | do you know what is available to load? | 09:37 |
gf712 | and then call the respective so when the factory is called? | 09:37 |
gf712 | sg.ASDF just has to map to a lazy function such as sg.machine('ASDF') that returns an error if it doesn't exist? | 09:38 |
@wiking | so say we have a plugin .so dir | 09:38 |
@wiking | and that's where you can put all your shit | 09:38 |
@wiking | and then when you do sg.machine('asdf') | 09:39 |
@wiking | then basically you try to load the libshogun-asdf.so | 09:39 |
@wiking | or whatever | 09:39 |
gf712 | yes, but that would work fine with this approach | 09:39 |
gf712 | at least in python/.. | 09:39 |
@wiking | yeah | 09:39 |
@wiking | but i'm saying that it doesn't work imo in jvm | 09:39 |
@wiking | dunno about ruby, .net | 09:39 |
@wiking | i mean we can add this convenience thing for sure | 09:40 |
gf712 | maybe could start a feature branch to test out a bit? | 09:40 |
@wiking | sure | 09:40 |
gf712 | just wanted to know if it would be worth trying out | 09:40 |
gf712 | wiking: anyway, the LibLinearRegression fails sometimes in MKL, sometimes in Accelerate but never for both :( | 09:41 |
gf712 | I have no idea what is happening | 09:42 |
gf712 | but it's just for lr_without_bias | 09:42 |
gf712 | which is very odd | 09:42 |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Remote host closed the connection] | 09:43 | |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 09:54 | |
-!- mode/#shogun [+o wiking] by ChanServ | 09:54 | |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Ping timeout: 240 seconds] | 09:58 | |
-!- wiking [~wiking@2001:67c:10ec:5784:8000::1153] has joined #shogun | 10:01 | |
-!- wiking [~wiking@2001:67c:10ec:5784:8000::1153] has quit [Changing host] | 10:01 | |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 10:01 | |
-!- mode/#shogun [+o wiking] by ChanServ | 10:01 | |
-!- Lefteris [~lefteris@2a05:b400:100:100:9013:13cd:ae89:b44] has joined #shogun | 10:32 | |
-!- glitch [~saphire@42.108.33.74] has joined #shogun | 10:57 | |
-!- glitch404 [uid346916@gateway/web/irccloud.com/x-abquhnmbdyvfhbve] has joined #shogun | 10:58 | |
glitch | hi this is glitch . | 10:58 |
glitch | I was looking forward for issue #3513 | 10:59 |
glitch | https://github.com/shogun-toolbox/shogun/issues/3513 | 10:59 |
glitch | would doing a documentation task put me away from developing team ? | 10:59 |
@wiking | ? | 11:03 |
@wiking | have you read the getting started guide? | 11:03 |
-!- ashwath [0e8be4dc@gateway/web/freenode/ip.14.139.228.220] has quit [Ping timeout: 256 seconds] | 11:03 | |
glitch | Yes , I have read through the guide | 11:06 |
glitch | https://github.com/shogun-toolbox/shogun/wiki/Google-Summer-of-Code-2019-Projects | 11:06 |
glitch | https://github.com/shogun-toolbox/shogun/wiki/Getting-involved | 11:06 |
glitch | getting involved | 11:06 |
glitch | and have successfully been through the building process | 11:06 |
glitch | and there , i found links to "entrance tasks" | 11:07 |
@wiking | k | 11:07 |
@wiking | so then just start sending in prs | 11:08 |
@wiking | doesn't matter the type of pr | 11:08 |
@wiking | of course on the end of the day for a successful application | 11:08 |
@wiking | we expect that the student would do coding | 11:08 |
@wiking | but in the beginning any sort of pr is fine | 11:08 |
@wiking | this gets your familiar with the dev cycle of shogun | 11:08 |
@wiking | (or any other open source project) | 11:08 |
glitch | thank you for the guidance , sir ! | 11:08 |
@wiking | it's all good and drop the 'sir' ;) | 11:09 |
@wiking | welcome to shogun | 11:09 |
glitch | :) thank you @wiking | 11:10 |
glitch | I will send a PR draft ASAP | 11:11 |
@wiking | sure thing | 11:11 |
glitch | for issue #3513 | 11:11 |
@wiking | if you are unsure, plz send a pr draft | 11:11 |
glitch | :D | 11:11 |
@wiking | as that doesn't eat up CI resources | 11:11 |
@wiking | ta | 11:11 |
glitch | sure | 11:11 |
gf712 | wiking: the python version of what I meant https://github.com/gf712/shogun/commit/3ac32e07786847ae8e573714ead8dc1165853191#diff-2ede860129e5a4430fa32403a0c6fffeR1472 | 11:11 |
-!- Lefteris [~lefteris@2a05:b400:100:100:9013:13cd:ae89:b44] has quit [Quit: Leaving] | 11:55 | |
-!- Ash__ [uid350320@gateway/web/irccloud.com/x-tlzfwozzeryogxet] has quit [Quit: Connection closed for inactivity] | 12:20 | |
-!- glitch [~saphire@42.108.33.74] has quit [Ping timeout: 245 seconds] | 12:27 | |
-!- trancenoid [0e8becd3@gateway/web/freenode/ip.14.139.236.211] has joined #shogun | 12:30 | |
-!- trancenoid [0e8becd3@gateway/web/freenode/ip.14.139.236.211] has quit [Ping timeout: 256 seconds] | 13:29 | |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Remote host closed the connection] | 13:58 | |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 13:59 | |
-!- mode/#shogun [+o wiking] by ChanServ | 13:59 | |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Remote host closed the connection] | 14:16 | |
-!- wiking [~wiking@2001:67c:10ec:5784:8000::1153] has joined #shogun | 14:16 | |
-!- wiking [~wiking@2001:67c:10ec:5784:8000::1153] has quit [Changing host] | 14:16 | |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 14:16 | |
-!- mode/#shogun [+o wiking] by ChanServ | 14:16 | |
-!- besser82 [~besser82@fedora/besser82] has quit [Quit: Freedom, Friends, Features, First [fedoraproject.org]] | 14:48 | |
-!- besser82 [~besser82@fedora/besser82] has joined #shogun | 14:50 | |
-!- mode/#shogun [+o besser82] by ChanServ | 14:50 | |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Remote host closed the connection] | 15:02 | |
-!- glitch404 [uid346916@gateway/web/irccloud.com/x-abquhnmbdyvfhbve] has quit [Quit: Connection closed for inactivity] | 15:08 | |
-!- wiking [~wiking@2001:67c:10ec:5784:8000::1153] has joined #shogun | 15:17 | |
-!- wiking [~wiking@2001:67c:10ec:5784:8000::1153] has quit [Changing host] | 15:17 | |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 15:17 | |
-!- mode/#shogun [+o wiking] by ChanServ | 15:17 | |
@wiking | anybody has any idea about localisation on windowser? | 15:20 |
@wiking | namely why the hack the very same thing that works on my windows 7 (reading file content) doesn't work on the windows at azure ci :) | 15:23 |
@wiking | gf712: :+1: | 15:23 |
@wiking | gf712: https://twitter.com/CompilerCpp | 15:33 |
gf712 | wiking: haha it'll be quiet until msvc starts pulling its weight... | 15:34 |
@wiking | gf712: dont be harsh on msvc | 15:35 |
@wiking | parallel stl is only available in msvc | 15:35 |
@wiking | ;) | 15:35 |
@wiking | https://devblogs.microsoft.com/cppblog/using-c17-parallel-algorithms-for-better-performance/ | 15:35 |
gf712 | yes, but I like argument packs and stuff... | 15:35 |
gf712 | and it doesn't do it :( | 15:36 |
@wiking | YOLO YOLO | 15:36 |
@wiking | :) | 15:36 |
@wiking | imagine that people still code in C98 standard | 15:36 |
@wiking | ;)( | 15:36 |
gf712 | I know... sometimes you just have to though | 15:37 |
gf712 | I guess parallel is better then | 15:37 |
gf712 | because packs you can replicate | 15:37 |
@wiking | but yeah | 15:37 |
@wiking | i have no clue | 15:37 |
@wiking | what's happening on azuer ci windows | 15:37 |
@wiking | :S | 15:37 |
gf712 | ah yes let me check | 15:37 |
@wiking | i mean | 15:37 |
@wiking | it cannot read the file | 15:37 |
@wiking | the very same stuff on my windows | 15:37 |
@wiking | works | 15:37 |
@wiking | it's not that the file doesn't exist | 15:37 |
@wiking | it complains about reading content | 15:38 |
gf712 | the 0xc0000409? | 15:38 |
@wiking | yep yep | 15:38 |
gf712 | can you give yourself sudo? | 15:38 |
@wiking | could not read array... | 15:38 |
@wiking | nono but you see | 15:38 |
@wiking | it opens the file | 15:38 |
gf712 | ah | 15:38 |
@wiking | if it couldn't open or there would be problem with opening | 15:39 |
@wiking | it would have a different error | 15:39 |
@wiking | it would throw a 'Error opening file' | 15:39 |
@wiking | when there's a problem wiht opening the file | 15:39 |
gf712 | but the reading operation returns a nullptr right? | 15:41 |
gf712 | as in the buffer is empty? | 15:41 |
@wiking | CSerializableFile::false_warn being called | 15:41 |
@wiking | which is when the read operation returns with a false | 15:42 |
gf712 | but it's false when it fails to read no? | 15:44 |
gf712 | when fscan fails? | 15:44 |
@wiking | yes | 15:45 |
@wiking | but i mean the funniest as said | 15:48 |
@wiking | other windowzer works | 15:48 |
@wiking | :> | 15:48 |
gf712 | hmmm very weird | 15:48 |
gf712 | is there a check for the path in lib shogun? | 15:48 |
gf712 | to see if the file exists? | 15:48 |
gf712 | and if a file has been written? | 15:49 |
gf712 | I.e. might be an issue with the saving operation? | 15:49 |
@wiking | it can write it :) | 15:49 |
@wiking | we are trying to read the reference txt | 15:50 |
@wiking | from data submodule | 15:50 |
@wiking | as said | 15:50 |
@wiking | FILE* fstream = fopen(fname, mode); | 15:50 |
@wiking | if (!fstream) { | 15:50 |
@wiking | this should catch file opening issues | 15:50 |
gf712 | ah ok, I didn't see that | 15:50 |
@wiking | so it doesn't complain about writing files | 15:51 |
@wiking | i.e. opens writes etc | 15:51 |
@wiking | as you can see the generated- tests are passing | 15:51 |
@wiking | so one presumes that those files are generated | 15:51 |
@wiking | and based on that | 15:52 |
@wiking | it can even read those files | 15:53 |
@wiking | as it doesn't complain at that bit | 15:53 |
gf712 | hmmm ok makes sense | 15:53 |
gf712 | so basically the files it is trying to read are empty? | 15:53 |
@wiking | it complains after | 15:53 |
@wiking | good q but something like that it seems | 15:53 |
gf712 | there wouldn't be encoding issues right? | 15:54 |
@wiking | good q | 15:54 |
@wiking | that was my idea as well | 15:54 |
@wiking | coz otherwise i have no clue why else would it work on another windows | 15:55 |
@wiking | but we even do a setlocale(LC_ALL, "C") | 15:56 |
@wiking | :) | 15:56 |
gf712 | https://en.cppreference.com/w/cpp/io/c/fopen | 15:56 |
gf712 | ah no wait | 15:56 |
gf712 | I was going to say UB but no | 15:57 |
gf712 | The behavior is undefined if the mode is not one of the strings listed above. Some implementations define additional supported modes (e.g. Windows). | 15:57 |
@wiking | "r" | 15:57 |
gf712 | yes I know :( | 15:58 |
gf712 | but in windows you can specify encoding https://docs.microsoft.com/en-us/cpp/c-runtime-library/reference/fopen-wfopen?view=vs-2017 | 15:58 |
@wiking | we have ascii | 16:00 |
@wiking | none of those should matter | 16:00 |
@wiking | LC_ALL = C should clear all these mumojumbo | 16:00 |
gf712 | hmmm | 16:02 |
gf712 | then the files are somehow empty | 16:03 |
gf712 | might be worth doing a cat/dir? | 16:03 |
@wiking | yeah i guess i have no other choice | 16:04 |
@wiking | type? | 16:04 |
@wiking | :) | 16:04 |
gf712 | type? | 16:06 |
-!- ssk441 [31ce0691@gateway/web/freenode/ip.49.206.6.145] has joined #shogun | 16:07 | |
@wiking | that's the windows eq of cat | 16:07 |
@wiking | :) | 16:07 |
@wiking | based on that test | 16:08 |
@wiking | there's the file and the content is correct | 16:08 |
@wiking | :) | 16:08 |
ssk441 | Hi there guys! I was hoping someone could help me generate the python examples for shogun. I'm new to CMake and I don't quite get what to do as specified in EXAMPLES.md | 16:09 |
@wiking | ssk441: you can use shogun via conda | 16:09 |
@wiking | no need to compile it from source | 16:09 |
@wiking | unless u wanna develop | 16:09 |
ssk441 | Yep I intend to participate through GSoC :D | 16:10 |
@wiking | then read the manual for how to get started | 16:10 |
@wiking | that should help you how to start | 16:10 |
@wiking | gf712: so it's really no clue :) | 16:11 |
ssk441 | wiking: I did. That's what led me to trying to generate the examples so I could get a feel for shogun | 16:12 |
ssk441 | I ran cmake with -DBUILD_META_EXAMPLES=ON and then make meta_examples as well. I'm not sure what comes next | 16:13 |
gf712 | wiking: i have no idea right now | 16:14 |
gf712 | ssk441: have you had a look at these http://www.shogun-toolbox.org/examples/latest/index.html ? | 16:18 |
ssk441 | gf712: I have but that involves running the examples. In my case, I'm unable to generate the examples. | 16:19 |
ssk441 | Looking at this right now: https://github.com/shogun-toolbox/docs/blob/master/EXAMPLES.md . Not clear on how to generate the examples | 16:20 |
gf712 | ah ok, sorry I misunderstood | 16:21 |
gf712 | ssk441: are you compiling with interfaces? I.e. python? | 16:21 |
ssk441 | Yeah I've compiled with the python interface. I've checked and shogun.py is in the location its supposed to be. | 16:24 |
gf712 | did you check examples/meta/python in the build directory? | 16:25 |
gf712 | assuming you want to see the meta examples? | 16:25 |
-!- geektoni [c1cdd24a@gateway/web/freenode/ip.193.205.210.74] has quit [Quit: Page closed] | 16:29 | |
ssk441 | I do want to see them and the directory only has a CMakeLists.txt | 16:30 |
gf712 | ssk441 I mean in the build directory | 16:32 |
gf712 | so build/examples/meta/python | 16:33 |
-!- glitch [~saphire@14.139.209.3] has joined #shogun | 16:33 | |
ssk441 | Ahhh. So it was in the build directory. Thanks so much for your time! :D | 16:35 |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Remote host closed the connection] | 16:38 | |
gf712 | ssk441 no problem! | 16:40 |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 17:02 | |
-!- mode/#shogun [+o wiking] by ChanServ | 17:03 | |
-!- ssk441 [31ce0691@gateway/web/freenode/ip.49.206.6.145] has quit [Quit: Page closed] | 17:04 | |
-!- HeikoS [5534e66b@gateway/web/cgi-irc/kiwiirc.com/ip.85.52.230.107] has joined #shogun | 17:34 | |
@wiking | gf712: still here? | 17:38 |
-!- chaitanya [0e8b5206@gateway/web/freenode/ip.14.139.82.6] has joined #shogun | 17:42 | |
HeikoS | yoyo | 17:43 |
chaitanya | I want to do GSoC project for Shogun.ml, where should I start? | 17:44 |
@wiking | chaitanya: there's a nice manual about how to get started. check our wiki (shogun repo) | 17:46 |
chaitanya | @wiking can you provide link for it? | 17:49 |
gf712 | wiking: yes still here | 17:50 |
@wiking | chaitanya: :) so far everybody managed to find it | 17:50 |
@wiking | dont get me wrong, i dont mind giving it to you | 17:50 |
@wiking | https://github.com/shogun-toolbox/shogun/wiki/Google-Summer-of-Code-2019-Projects | 17:50 |
@wiking | https://github.com/shogun-toolbox/shogun/wiki/Getting-involved | 17:50 |
gf712 | HeikoS: hey, just got your email for ATI, I'll go through it tomorrow | 17:51 |
@wiking | but we expect much more autonomy from gsoc students | 17:51 |
HeikoS | ok sure nw | 17:51 |
@wiking | gf712: do u know a quite big cnn for obj recongnition? | 17:51 |
@wiking | wanna test some coreml stuff | 17:51 |
HeikoS | wiking gf712 did you see the gradient refactoring PR? | 17:52 |
gf712 | vgg16 should be as big as it gets? | 17:52 |
@wiking | HeikoS: not yet | 17:52 |
gf712 | or do you want stuff for bounding box | 17:52 |
@wiking | gf712: bounding box | 17:52 |
@wiking | and not yolo | 17:52 |
gf712 | YOLO | 17:52 |
gf712 | ah | 17:52 |
@wiking | i mean not yolov3 | 17:52 |
@wiking | thnx | 17:52 |
@wiking | :D | 17:52 |
@wiking | that code | 17:52 |
@wiking | maaana | 17:52 |
@wiking | have u read the code? | 17:52 |
@wiking | :D | 17:52 |
@wiking | because i did | 17:52 |
@wiking | :D | 17:52 |
gf712 | HeikoS: I went through it briefly | 17:53 |
@wiking | void* x = calloc(...) | 17:53 |
@wiking | x[i] = | 17:53 |
@wiking | return value check | 17:53 |
@wiking | fuck it | 17:53 |
@wiking | :D | 17:53 |
gf712 | wiking: I did, I implemented it in tf once | 17:53 |
@wiking | (and of course i ran out of memory) | 17:53 |
gf712 | to test distributed GPU stuff | 17:53 |
@wiking | so calloc actually returned nullptr | 17:53 |
gf712 | next biggest would be the streetview ones I guess? | 17:54 |
@wiking | ok vgg16 should be good | 17:54 |
@wiking | lol | 17:54 |
@wiking | good start | 17:54 |
@wiking | protobuf 556 megz | 17:54 |
@wiking | :D | 17:54 |
@wiking | nice one | 17:55 |
gf712 | wiking checkout coco | 17:55 |
@wiking | lemme test | 17:55 |
@wiking | gf712: because i'm just writing the mmap part of the coreml engine | 17:55 |
gf712 | it's the most famous dataset | 17:55 |
HeikoS | wiking gf712 if you have a minute, check out how gradients work in shogun (any gp example/test will do). And then have a look at the PR by Lefteris. He refactored the old approach to the new param framework. But I have doubts about the whole design tbh, so before putting in all this work, we could maybe make that a bit more clean | 17:55 |
gf712 | ah ok | 17:55 |
@wiking | since i realised that i'm having trouble running this on embedded device | 17:55 |
@wiking | where this is the ram | 17:56 |
@wiking | Mem: 494M 74M 99M 5.5M 320M 195M | 17:56 |
@wiking | 195M = availalbe | 17:56 |
gf712 | wiking: lol | 17:56 |
gf712 | HeikoS: I will have a better look now | 17:56 |
gf712 | I am not a big fan of adding name to AnyParameters btw | 17:56 |
gf712 | it's a bit redundant | 17:56 |
HeikoS | no me neither | 17:56 |
HeikoS | well it was in TParameter before | 17:57 |
HeikoS | but that's partly why I asked you to think about the gradients | 17:57 |
HeikoS | I think we can do a bit better | 17:57 |
@wiking | gf712: the thing is that it'd be ggreat to have coreml as part of tracker (https://wiki.gnome.org/Projects/Tracker | 17:57 |
@wiking | but then this should work on NAS and shit like this | 17:57 |
HeikoS | we just need a way for shogun devs to manually implement gradients of models wrt to certain parameters | 17:57 |
HeikoS | so some dev-API | 17:57 |
@wiking | gf712: that ram is a linux nas | 17:58 |
gf712 | HeikoS: I'll see what Lefteris wrote and see if I can come up with better stuff | 17:59 |
HeikoS | i think it might be good to think about how to do it before getting into it too much | 18:00 |
HeikoS | as it is a big pr | 18:00 |
HeikoS | but yeah would be good to get some input | 18:00 |
gf712 | wiking: the tracker thing seems cool | 18:00 |
HeikoS | wiking gf712 are you fine with merging https://github.com/shogun-toolbox/shogun/pull/4555? | 18:00 |
gf712 | wiking: out of RAM yet? | 18:00 |
HeikoS | only mkl mac fails but I guess that is unrelated | 18:00 |
@wiking | mergeit | 18:02 |
@wiking | we have a regression | 18:02 |
gf712 | HeikoS: the PR seems fine | 18:03 |
HeikoS | ok! | 18:03 |
gf712 | HeikoS: the CI issues are due to LibLinearRegression | 18:04 |
HeikoS | i see | 18:05 |
gf712 | I am not sure why as it only fails sometimes | 18:05 |
gf712 | and it can be either with mkl or accelerate | 18:05 |
gf712 | but there are only issues on clang | 18:05 |
gf712 | I can reproduce it locally if run the test several times it will fail occasionally | 18:05 |
HeikoS | weird | 18:10 |
gf712 | HeikoS I tried setting a seed in the test but it didn't help | 18:11 |
gf712 | also it's only the test without bias that sometimes fails | 18:11 |
HeikoS | oh my | 18:13 |
HeikoS | I guess would need to check what code it uses under the hood | 18:13 |
gf712 | I've checked for issues in initialisation, but couldn't find any problems | 18:15 |
gf712 | everything is set to 0 | 18:15 |
HeikoS | gf712 btw did you register for the ws in the end? | 18:16 |
HeikoS | just got an email that it is last day to register on eventb | 18:16 |
gf712 | really? | 18:16 |
gf712 | they told me it was full... | 18:16 |
gf712 | can you send me a link? | 18:16 |
gf712 | I'll register now then | 18:17 |
-!- glitch [~saphire@14.139.209.3] has quit [Ping timeout: 268 seconds] | 18:17 | |
HeikoS | sent | 18:20 |
HeikoS | gf712 ok I gotta bounce now, let me know your thoughts on the gradient stuff so we can get back to Lefteris | 18:20 |
gf712 | HeikoS: thanks for the email | 18:21 |
gf712 | but it doesn't have the link | 18:21 |
gf712 | HeikoS: the link seems to have gotten lost? | 18:21 |
HeikoS | sent another one | 18:22 |
HeikoS | they told me to not share the link | 18:22 |
HeikoS | but I think you should fill it anyways :) | 18:22 |
HeikoS | since you are just there one day | 18:22 |
gf712 | HeikoS: thanks! | 18:23 |
gf712 | I'll register now | 18:23 |
HeikoS | cool | 18:24 |
HeikoS | btw Ill also start making a poster soon | 18:25 |
HeikoS | let me know if you have comments on that | 18:25 |
HeikoS | or ideas | 18:25 |
HeikoS | as i will start from scratch | 18:25 |
-!- glitch [~saphire@14.139.209.3] has joined #shogun | 18:29 | |
HeikoS | ok signing off, see you gf712! | 18:30 |
gf712 | HeikoS OK! | 18:31 |
gf712 | ttyl | 18:31 |
-!- HeikoS [5534e66b@gateway/web/cgi-irc/kiwiirc.com/ip.85.52.230.107] has quit [Ping timeout: 245 seconds] | 18:34 | |
-!- gf712 [c13cdcfd@gateway/web/freenode/ip.193.60.220.253] has quit [Ping timeout: 256 seconds] | 18:35 | |
-!- chaitanya [0e8b5206@gateway/web/freenode/ip.14.139.82.6] has quit [Ping timeout: 256 seconds] | 18:56 | |
-!- glitch [~saphire@14.139.209.3] has quit [Ping timeout: 255 seconds] | 19:00 | |
@wiking | [libprotobuf ERROR google/protobuf/io/zero_copy_stream_impl_lite.cc:164] Cannot allocate buffer larger than kint32max for StringOutputStream. | 19:01 |
@wiking | lOOOOL | 19:01 |
-!- vi-to [~Vito@93-40-66-36.ip37.fastwebnet.it] has joined #shogun | 20:13 | |
vi-to | hi :) I am having issues compiling shogun. It is not a shogun issue, so I asked on stackoverflow (I asked on stackoverflow (https://stackoverflow.com/questions/19915683/how-to-find-package-for-installed-file-in-brew) | 20:15 |
vi-to | but maybe someone of you experienced the same problem and solved it | 20:17 |
-!- vi-to [~Vito@93-40-66-36.ip37.fastwebnet.it] has left #shogun [] | 20:26 | |
-!- kartik [31248371@gateway/web/freenode/ip.49.36.131.113] has joined #shogun | 20:33 | |
-!- kartik [31248371@gateway/web/freenode/ip.49.36.131.113] has quit [Client Quit] | 20:35 | |
-!- vi-to [~Vito@93-40-66-36.ip37.fastwebnet.it] has joined #shogun | 20:36 | |
-!- RUPESH [671b0868@gateway/web/freenode/ip.103.27.8.104] has joined #shogun | 20:44 | |
-!- RUPESH [671b0868@gateway/web/freenode/ip.103.27.8.104] has quit [Quit: Page closed] | 21:28 | |
-!- vi-to [~Vito@93-40-66-36.ip37.fastwebnet.it] has left #shogun [] | 23:15 | |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Remote host closed the connection] | 23:29 | |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 23:30 | |
-!- mode/#shogun [+o wiking] by ChanServ | 23:30 | |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Ping timeout: 255 seconds] | 23:36 | |
--- Log closed Thu Mar 07 00:00:40 2019 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!