--- Log opened Tue Jun 14 00:00:39 2016 | ||
-!- sonne|osx [~sonne@x4e31fd0a.dyn.telefonica.de] has quit [Quit: sonne|osx] | 00:17 | |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 01:04 | |
shogun-notifier- | shogun: Ariane Paola Gomes :develop * 2aab17c / examples/meta/generator/generate.py: https://github.com/shogun-toolbox/shogun/commit/2aab17cc7cf099dfa3e8ffbc8d361d3ebd000f25 | 01:04 |
---|---|---|
shogun-notifier- | shogun: Fixes #3253 Cookbook generation failure missing responsible file indicator. | 01:04 |
shogun-notifier- | shogun: Heiko Strathmann :develop * 997e54c / examples/meta/generator/generate.py: https://github.com/shogun-toolbox/shogun/commit/997e54c7a262baf7833454f154593398bb935ff3 | 01:04 |
shogun-notifier- | shogun: Merge pull request #3255 from arianepaola/fix_3253 | 01:04 |
shogun-notifier- | shogun: | 01:04 |
shogun-notifier- | shogun: Fixes #3253 Cookbook generation failure missing file indicator | 01:04 |
shogun-buildbot | build #2912 of bsd1 - libshogun is complete: Failure [failed configure] Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/2912 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Ariane Paola Gomes <arianepaola@users.noreply.github.com> | 01:12 |
shogun-buildbot | build #53 of xenial - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/xenial%20-%20libshogun/builds/53 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Ariane Paola Gomes <arianepaola@users.noreply.github.com> | 01:13 |
shogun-buildbot | build #881 of deb4 - python3 is complete: Failure [failed install test python modular] Build details are at http://buildbot.shogun-toolbox.org/builders/deb4%20-%20python3/builds/881 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Ariane Paola Gomes <arianepaola@users.noreply.github.com> | 01:40 |
shogun-buildbot | build #41 of FC23 - libshogun - aarch64 is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/FC23%20-%20libshogun%20-%20aarch64/builds/41 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Ariane Paola Gomes <arianepaola@users.noreply.github.com> | 01:49 |
-!- besser82_ [~besser82@fedora/besser82] has joined #shogun | 01:54 | |
-!- mode/#shogun [+o besser82_] by ChanServ | 01:54 | |
-!- besser82_ [~besser82@fedora/besser82] has quit [Ping timeout: 260 seconds] | 02:16 | |
shogun-buildbot | build #1020 of nightly_none is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/nightly_none/builds/1020 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Ariane Paola Gomes <arianepaola@users.noreply.github.com> | 03:04 |
-!- GandalfTheWizard [~Eva@112.10.170.90] has joined #shogun | 03:07 | |
-!- OXPHOS [9d8b131c@gateway/web/freenode/ip.157.139.19.28] has quit [Quit: Page closed] | 04:01 | |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 04:04 | |
shogun-buildbot | build #19 of memleak - valgrind is complete: Failure [failed memory check] Build details are at http://buildbot.shogun-toolbox.org/builders/memleak%20-%20valgrind/builds/19 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Ariane Paola Gomes <arianepaola@users.noreply.github.com> | 05:28 |
-!- mizari [~mizari@isc.tsu.ru] has joined #shogun | 05:44 | |
shogun-buildbot | build #1150 of nightly_default is complete: Failure [failed test notebooks] Build details are at http://buildbot.shogun-toolbox.org/builders/nightly_default/builds/1150 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Ariane Paola Gomes <arianepaola@users.noreply.github.com> | 06:10 |
-!- mizari [~mizari@isc.tsu.ru] has quit [Ping timeout: 244 seconds] | 06:50 | |
shogun-buildbot | build #470 of deb1 - libshogun - PR is complete: Failure [failed cookbook] Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun%20-%20PR/builds/470 blamelist: OXPHOS | 07:13 |
shogun-buildbot | build #471 of deb1 - libshogun - PR is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun%20-%20PR/builds/471 | 07:17 |
-!- dennyjooo [~dennyjooo@223.206.166.254] has joined #shogun | 08:25 | |
-!- mizari [~mizari@isc.tsu.ru] has joined #shogun | 08:31 | |
-!- mizari [~mizari@isc.tsu.ru] has quit [Quit: Leaving] | 08:36 | |
-!- dennyjooo [~dennyjooo@223.206.166.254] has quit [] | 08:55 | |
-!- dennyjoo_ [~dennyjooo@223.206.166.254] has joined #shogun | 08:56 | |
-!- dennyjoo_ [~dennyjooo@223.206.166.254] has quit [] | 09:08 | |
-!- lambday [0551433c@gateway/web/freenode/ip.5.81.67.60] has quit [Quit: Page closed] | 09:08 | |
@wiking | Saurabh7: hey here? | 10:03 |
Saurabh7 | wiking: yes, sry was away | 10:23 |
-!- besser82_ [~besser82@fedora/besser82] has joined #shogun | 10:26 | |
-!- mode/#shogun [+o besser82_] by ChanServ | 10:26 | |
Saurabh7 | I have updated things, I will ping there | 10:30 |
@wiking | Saurabh7: yep i saw them | 10:34 |
@wiking | i was just wondering if you maybe benchmarked it | 10:35 |
@wiking | after the final changes | 10:35 |
@wiking | (lars) | 10:35 |
Saurabh7 | Yes same results not much difference | 10:36 |
Saurabh7 | but I havent changed much that has impact | 10:36 |
Saurabh7 | just chcanged memcpy for 1 column | 10:37 |
-!- ptizoom [~christian@host-92-21-196-52.as13285.net] has joined #shogun | 10:37 | |
Saurabh7 | wiking: ^ | 10:38 |
@wiking | Saurabh7: perfect | 10:41 |
@wiking | ok let's see the output of travis | 10:41 |
@wiking | if it's green we merge | 10:41 |
@wiking | :) | 10:41 |
@wiking | ok | 10:42 |
@wiking | kmeans is grean | 10:42 |
@wiking | *green | 10:43 |
@wiking | mergeoooooo | 10:43 |
Saurabh7 | cool :) | 10:43 |
@wiking | boom | 10:44 |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 10:44 | |
shogun-notifier- | shogun: Saurabh7 :develop * 802b810 / / (13 files): https://github.com/shogun-toolbox/shogun/commit/802b8100080bd28e1099950176d69c437ad29e05 | 10:44 |
shogun-notifier- | shogun: refactor KMeans | 10:44 |
shogun-notifier- | shogun: Viktor Gal :develop * e541a92 / / (13 files): https://github.com/shogun-toolbox/shogun/commit/e541a9288a12c2ee1b70f7944feae2246f178bea | 10:44 |
shogun-notifier- | shogun: Merge pull request #3166 from Saurabh7/kmeans_refactor | 10:44 |
shogun-notifier- | shogun: | 10:44 |
shogun-notifier- | shogun: refactor KMeans | 10:44 |
Saurabh7 | aha | 10:44 |
Saurabh7 | oh theres a kmeans cookbook open | 10:49 |
Saurabh7 | guess it needs updating :) | 10:50 |
shogun-buildbot | build #2913 of bsd1 - libshogun is complete: Failure [failed configure] Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/2913 blamelist: Viktor Gal <vigsterkr@gmail.com>, Saurabh7 <saurabh.mahindre@gmail.com> | 10:51 |
-!- besser82_ [~besser82@fedora/besser82] has quit [Ping timeout: 272 seconds] | 10:54 | |
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun | 10:54 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 10:54 | |
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has quit [Ping timeout: 258 seconds] | 11:02 | |
-!- lambday [8028b10a@gateway/web/freenode/ip.128.40.177.10] has joined #shogun | 11:28 | |
-!- mode/#shogun [+o lambday] by ChanServ | 11:28 | |
-!- c4goldsw [5da420e6@gateway/web/cgi-irc/kiwiirc.com/ip.93.164.32.230] has joined #shogun | 11:29 | |
c4goldsw | wiking ping | 11:29 |
shogun-buildbot | build #882 of deb4 - python3 is complete: Failure [failed install test python modular] Build details are at http://buildbot.shogun-toolbox.org/builders/deb4%20-%20python3/builds/882 blamelist: Viktor Gal <vigsterkr@gmail.com>, Saurabh7 <saurabh.mahindre@gmail.com> | 11:37 |
-!- HeikoS [~heiko@nat-202-76.internal.eduroam.ucl.ac.uk] has joined #shogun | 12:07 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 12:07 | |
c4goldsw | clear | 12:37 |
c4goldsw | oops, meant for bash | 12:40 |
-!- besser82_ [~besser82@fedora/besser82] has joined #shogun | 12:41 | |
-!- mode/#shogun [+o besser82_] by ChanServ | 12:41 | |
@wiking | c4goldsw: pong | 12:45 |
c4goldsw | wiking hey | 12:46 |
c4goldsw | doing well? | 12:46 |
@wiking | yep yep | 12:46 |
@wiking | how about you? | 12:46 |
c4goldsw | Good - I was at an ML hackathon in Germany this weekend, so I haven't done much on Shogun | 12:47 |
c4goldsw | getting back to work on it. | 12:47 |
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Quit: buildmaster reconfigured: bot disconnecting] | 12:49 | |
c4goldsw | afk | 12:54 |
-!- besser82_ [~besser82@fedora/besser82] has quit [Ping timeout: 250 seconds] | 13:01 | |
-!- HeikoS [~heiko@nat-202-76.internal.eduroam.ucl.ac.uk] has quit [Quit: Leaving.] | 13:11 | |
-!- c4goldsw [5da420e6@gateway/web/cgi-irc/kiwiirc.com/ip.93.164.32.230] has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client] | 13:18 | |
-!- c4goldsw [5da420e6@gateway/web/cgi-irc/kiwiirc.com/ip.93.164.32.230] has joined #shogun | 13:19 | |
-!- GandalfTheWizard [~Eva@112.10.170.90] has quit [Quit: Leaving.] | 13:24 | |
c4goldsw | wiking question: I started off by making EuclideanDistance generic, which would also involve making it's parent class generic (which would then involve making all descendents of the base class generic). So, at this point, it seems as if I'd have to do this for all distance classes (I count + 15). I should try to find something with fewer depende | 13:42 |
c4goldsw | ncies and start over, right? | 13:42 |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 13:44 | |
c4goldsw | Unless you wouldn't mind going through all those files. | 13:49 |
c4goldsw | In which case I'd keep going | 13:49 |
@wiking | c4goldsw: mmmm | 13:51 |
@wiking | lemme checkin | 13:51 |
@wiking | ah yeah that's nasty | 13:52 |
@wiking | lemme find you an easier | 13:53 |
@wiking | check LARS | 13:53 |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 13:53 | |
shogun-notifier- | shogun: Saurabh7 :develop * 5aa89b7 / / (3 files): https://github.com/shogun-toolbox/shogun/commit/5aa89b7d1b3f59de03bb98b0f5ea90ea114f12d9 | 13:53 |
shogun-notifier- | shogun: Update LARS | 13:53 |
shogun-notifier- | shogun: Viktor Gal :develop * 3618469 / / (3 files): https://github.com/shogun-toolbox/shogun/commit/3618469f489698e6ac15d6e3a6938d3210be7b80 | 13:53 |
shogun-notifier- | shogun: Merge pull request #3243 from Saurabh7/lars | 13:53 |
shogun-notifier- | shogun: | 13:53 |
shogun-notifier- | shogun: Lars update | 13:53 |
@wiking | just merged some updates there | 13:53 |
@wiking | but that thing prettey much assumes the same thing | 13:54 |
@wiking | that the features is actually DenseFeatures<float64_t> | 13:54 |
@wiking | (only | 13:54 |
@wiking | which is for no reason :P | 13:54 |
c4goldsw | Haha, ok, I'll pull that then. I have a git question too | 13:55 |
c4goldsw | I have PR on my develop branch that is still open - if I want to start working on new work, should I make a new branch? | 13:55 |
c4goldsw | And work from there? | 13:56 |
@wiking | either that or just recreate the branch | 13:58 |
@wiking | delete and create | 13:58 |
@wiking | :) | 13:58 |
c4goldsw | Which would also delete the PR, right? | 13:58 |
@wiking | noup | 13:59 |
@wiking | but no reason to delete the pr | 13:59 |
@wiking | the pr will just monitor that branch | 13:59 |
c4goldsw | Great, thanks for the info. | 13:59 |
@wiking | which i suppose u wanna have | 13:59 |
@wiking | if not | 13:59 |
@wiking | then just close the pr via github | 13:59 |
c4goldsw | I'll keep it open, just want to make sure it isn't affected. | 14:00 |
@wiking | mmm buildbot is sick :( | 14:10 |
c4goldsw | wiking: Since LARS inherits from CLinearMachine, which inherits from CMachine, should I just make the methods in LARs generic rather than the whole class? | 14:10 |
@wiking | what exactly do you mean | 14:12 |
@wiking | i mean the problem in lars is this exactly | 14:12 |
@wiking | aaaaaaaaaaaaaaaaaaah | 14:13 |
@wiking | Saurabh7: still here? | 14:13 |
@wiking | c4goldsw: https://github.com/shogun-toolbox/shogun/blob/develop/src/shogun/regression/LeastAngleRegression.cpp#L118 | 14:13 |
c4goldsw | Sure, I have the line open in front of me. | 14:14 |
@wiking | Saurabh7: you've forgot this here: https://github.com/shogun-toolbox/shogun/blob/develop/src/shogun/regression/LeastAngleRegression.cpp#L13 | 14:14 |
@wiking | Saurabh7: you've removed all the lapack dependencies no? | 14:14 |
@wiking | because then i reckon we should remove https://github.com/shogun-toolbox/shogun/blob/develop/src/shogun/regression/LeastAngleRegression.cpp#L13 and https://github.com/shogun-toolbox/shogun/blob/develop/src/shogun/regression/LeastAngleRegression.cpp#L406 | 14:14 |
c4goldsw | wiking you know what, never mind. I just realised that there shouldn't be a problem with making the whole class generic. | 14:18 |
c4goldsw | (or so I think) | 14:18 |
@wiking | c4goldsw: hehe cool :) | 14:19 |
arianepaola | hello everyone :-) | 14:33 |
c4goldsw | wiking sorry to keep bother you, but I merged the main repo with my fork and it isn't compiling. It fails at "Generating examples from meta-language (I couldn't think of any solutions). Here's the output: https://gist.github.com/c4goldsw/2b7d1597d13472f7dd3bd92b014ed6ea | 14:34 |
c4goldsw | arianepaola: Hello | 14:34 |
arianepaola | hi c4goldsw | 14:34 |
@wiking | mmm good question | 14:35 |
@wiking | lol | 14:35 |
@wiking | try to remove the build directory | 14:35 |
@wiking | and recompile | 14:35 |
@wiking | btw | 14:36 |
@wiking | install ccache if you dont have it yet | 14:36 |
c4goldsw | Does it just keep track of what bytecode has / hasn't changed? | 14:36 |
arianepaola | hi wiking | 14:36 |
@wiking | hihi1 | 14:36 |
@wiking | c4goldsw: yep should | 14:36 |
c4goldsw | Sounds useful. Let's hope rebuilding works lol. | 14:37 |
arianepaola | wiking: the PR for setup.py is squashed | 14:37 |
@wiking | arianepaola: i'll try debugging now | 14:37 |
@wiking | but if i cannot find it in the next 1-1.5 hours | 14:37 |
@wiking | i'll just merge it | 14:37 |
arianepaola | ok, thanks | 14:37 |
c4goldsw | wiking still getting the same error after rebuilding from scratch - are there any issues on your end? | 15:08 |
@wiking | noup :) | 15:14 |
c4goldsw | In that case, I'll dig around myself | 15:16 |
c4goldsw | wiking ^ | 15:16 |
Saurabh7 | back | 15:41 |
Saurabh7 | wiking: there is an optional routine I had not touched, since no implciations | 15:42 |
Saurabh7 | it has some lapack i think | 15:42 |
Saurabh7 | If i replace it , then can remvoe the guards | 15:43 |
-!- [Chris] [~Chris]@deadtime.informatik.uni-tuebingen.de] has quit [Ping timeout: 276 seconds] | 15:44 | |
@wiking | Saurabh7: aaaah | 15:51 |
@wiking | would be difficult to replace that optional routine? | 15:51 |
@wiking | or just put that safeguard around the optional routine? | 15:52 |
@wiking | not on the whole class implementtion? | 15:52 |
Saurabh7 | I can replace I think | 15:53 |
Saurabh7 | yeah no point in guarding whole thing just for it | 15:54 |
@wiking | yeah so basically woudl be at least put the guard around the optional thingy | 15:57 |
-!- GandalfTheWizard [~Eva@112.10.170.90] has joined #shogun | 16:13 | |
Saurabh7 | hm whats this with python 3 except Exception, err: | 16:28 |
Saurabh7 | ^ | 16:28 |
Saurabh7 | SyntaxError: invalid syntax | 16:28 |
-!- OXPHOS [9d8b131c@gateway/web/freenode/ip.157.139.19.28] has joined #shogun | 16:34 | |
Saurabh7 | wiking: ok I removed the guards https://github.com/shogun-toolbox/shogun/pull/3288 | 16:36 |
@wiking | mmm where do u get this? | 16:37 |
Saurabh7 | develop | 16:37 |
Saurabh7 | building with meta on | 16:37 |
-!- HeikoS [~heiko@82-132-216-64.dab.02.net] has joined #shogun | 16:39 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 16:39 | |
shogun-notifier- | shogun: OXPHOS :develop * 02d277b / / (3 files): https://github.com/shogun-toolbox/shogun/commit/02d277b053fdd0ed0f30c15f47ca2adfa7305837 | 16:42 |
shogun-notifier- | shogun: cookbook - final | 16:42 |
shogun-notifier- | shogun: Heiko Strathmann :develop * f8155b1 / / (3 files): https://github.com/shogun-toolbox/shogun/commit/f8155b157f485bdbafa12cfaaabe4f2aedcce4bc | 16:42 |
shogun-notifier- | shogun: Merge pull request #3244 from OXPHOS/cookbook_mclr | 16:42 |
shogun-notifier- | shogun: | 16:42 |
shogun-notifier- | shogun: cookbook: multiclass logistic regression | 16:42 |
shogun-notifier- | shogun-data: OXPHOS :master * 81a6441 / testsuite/meta/clustering/kmeans.dat: https://github.com/shogun-toolbox/shogun-data/commit/81a64416c3f91d2f5aa333c8dac9667d4a4db54c | 16:53 |
shogun-notifier- | shogun-data: dataset | 16:53 |
shogun-notifier- | shogun-data: Heiko Strathmann :master * e316fd6 / testsuite/meta/clustering/kmeans.dat: https://github.com/shogun-toolbox/shogun-data/commit/e316fd67f45cb6e1b8b799369bf50878ddb0c560 | 16:54 |
shogun-notifier- | shogun-data: Merge pull request #101 from OXPHOS/temp | 16:54 |
shogun-notifier- | shogun-data: | 16:54 |
shogun-notifier- | shogun-data: update kmeans test dataset | 16:54 |
OXPHOS | lambday: hey! | 16:57 |
OXPHOS | HeikoS: thx for the merge | 16:57 |
@lambday | OXPHOS: hey | 16:57 |
@HeikoS | OXPHOS: thanks for da patch! | 16:57 |
OXPHOS | HeikoS: np! | 16:58 |
OXPHOS | lambday: I was thinking, there will always be a CPUVector, and the transfertoGPU only makes a copy to GPU | 16:59 |
-!- HeikoS1 [~heiko@82-132-215-139.dab.02.net] has joined #shogun | 16:59 | |
OXPHOS | lambday: and actually the GPU vector is never changed. all the operations (now) return CPU type | 17:00 |
@lambday | OXPHOS: umm not sure I fully understand | 17:00 |
@lambday | OXPHOS: can you please elaborate a bit? | 17:00 |
OXPHOS | lambday: so GPUVec is just an add-on to the Vector once users call transfertoGPU. and transfertoGPU doesn't change CPUVector at all | 17:01 |
OXPHOS | So it won't happen that there's no CPUVector | 17:02 |
@lambday | OXPHOS: no it doesn't change the cpu vector | 17:02 |
shogun-notifier- | shogun: OXPHOS :develop * 36cf05c / / (5 files): https://github.com/shogun-toolbox/shogun/commit/36cf05c6e15c93b661167f606fee88a8f7aa14a2 | 17:02 |
shogun-notifier- | shogun: cookbook - final | 17:02 |
shogun-notifier- | shogun: Heiko Strathmann :develop * da02a92 / / (5 files): https://github.com/shogun-toolbox/shogun/commit/da02a9280aa2fea485be2fe670aaeff342324962 | 17:02 |
shogun-notifier- | shogun: Merge pull request #3280 from OXPHOS/cookbook_ecoc_random | 17:02 |
shogun-notifier- | shogun: | 17:02 |
@lambday | OXPHOS: oh so you wanted to understand why we need to transfer the gpu memory back to cpu ? | 17:02 |
shogun-notifier- | shogun: Cookbook ecoc random | 17:02 |
OXPHOS | lambday: actually yeah | 17:02 |
-!- HeikoS [~heiko@82-132-216-64.dab.02.net] has quit [Ping timeout: 260 seconds] | 17:02 | |
HeikoS1 | big improvement that was | 17:03 |
@lambday | OXPHOS: so let's imagine this scenario: you transfer the memory to gpu.. | 17:03 |
@lambday | now you perform some operation on the vector in-place.. | 17:03 |
@lambday | as in, say, all the elements are squared.. or something like that | 17:04 |
@lambday | since the data was on gpu, the gpubackend then changed the data on gpu as per the operation specifies it | 17:04 |
@lambday | but it didn't touch cpu data at all | 17:04 |
@lambday | so, at the end of all your calls, if you want to return your vector, then the content of the cpu memory is different (obsolete) than that of gpu mem array | 17:05 |
@lambday | so if you don't copy the gpu memory to cpu then, the result is incorrect! | 17:05 |
-!- HeikoS1 [~heiko@82-132-215-139.dab.02.net] has quit [Quit: Leaving.] | 17:06 | |
@lambday | because the user would expect that all he did in between should be reflected in the result, no? | 17:06 |
@lambday | as in, in-place element-wise square, etc etc | 17:06 |
@lambday | OXPHOS: am I making sense? | 17:06 |
OXPHOS | lambday: yes but i am thinking | 17:06 |
OXPHOS | only a few operations require in-place change | 17:07 |
@lambday | OXPHOS: would it be safe to assume that? | 17:07 |
OXPHOS | should we just return the cpu type from the operation? | 17:07 |
@lambday | OXPHOS: return from where? | 17:08 |
OXPHOS | lambday: I might have confused sth. - so say T dot() | 17:09 |
OXPHOS | the T is on GPU if ViennaCL is called? or doesn't matter | 17:09 |
@lambday | OXPHOS: dot is easy.. dot returns a single scalar value.. | 17:09 |
@lambday | OXPHOS: what about element-wise square/sin/cos/log/exp/pow(N) ? | 17:10 |
@lambday | I mean, in-place operations would be there sometimes I think | 17:10 |
OXPHOS | lambday: yeah u'r right.. | 17:10 |
@lambday | you don't want to create a new memory every-time you do a small operation.. sometimes it is better to be done in-place | 17:10 |
@lambday | OXPHOS: say, you want to add two vectors, a and b, and want to store the result in a itself.. | 17:11 |
@lambday | this is another example | 17:11 |
OXPHOS | lambday: sure! and for the transfer to CPU - if onGPU=false, we throw SG_SINFO? | 17:14 |
@lambday | so, in the end, when all is said and done, you transfer your gpu data to cpu and return that | 17:14 |
@lambday | OXPHOS: yeah, should not be an error.. | 17:14 |
OXPHOS | lambday: and the threadsafe, should it be sth. like mutex? how sure how to accomplish that? | 17:18 |
@lambday | OXPHOS: yeah something like mutex or atomic or so.. just wanted that you thought about this a bit | 17:19 |
@lambday | OXPHOS: doesn't have to be in this patch.. but maybe in next patch you can add that | 17:19 |
OXPHOS | lambday: cool thx! | 17:19 |
@lambday | OXPHOS: issues like, say, two threads want to transfer the data to gpu.. and such and such | 17:19 |
OXPHOS | I see | 17:21 |
arianepaola | HeikoS would it be possible to add the content from the wiki, containing the description to README.md? | 17:22 |
-!- c4goldsw [5da420e6@gateway/web/cgi-irc/kiwiirc.com/ip.93.164.32.230] has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client] | 17:56 | |
-!- HeikoS [~heiko@nat-204-128.internal.eduroam.ucl.ac.uk] has joined #shogun | 18:20 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 18:20 | |
@HeikoS | Saurabh7: hi! | 18:21 |
@HeikoS | arianepaola: ping! | 18:21 |
Saurabh7 | HeikoS: hihi | 18:23 |
shogun-notifier- | shogun-data: OXPHOS :master * 8bc728a / testsuite/meta/classifier/relaxed_tree.dat: https://github.com/shogun-toolbox/shogun-data/commit/8bc728a585c3457cd3480b81ef63d80974fb97ea | 18:23 |
shogun-notifier- | shogun-data: data | 18:23 |
shogun-notifier- | shogun-data: Heiko Strathmann :master * d4c420b / testsuite/meta/classifier/relaxed_tree.dat: https://github.com/shogun-toolbox/shogun-data/commit/d4c420b6c93f16b3c3439f45de7245d7f432ff04 | 18:23 |
shogun-notifier- | shogun-data: Merge pull request #100 from OXPHOS/master | 18:23 |
shogun-notifier- | shogun-data: | 18:23 |
shogun-notifier- | shogun-data: integration test dataset - relaxed tree classifier | 18:23 |
@HeikoS | Saurabh7: happy about the merge? | 18:23 |
@HeikoS | Saurabh7: lets discuss next steps | 18:23 |
@HeikoS | Saurabh7: I have a little side thing: | 18:23 |
Saurabh7 | HeikoS: yes ofc :0 | 18:23 |
Saurabh7 | :) | 18:23 |
@HeikoS | so xvalidation | 18:23 |
@HeikoS | we have that parfor thing | 18:23 |
@HeikoS | what happens if I use parfor in an algorithm? | 18:24 |
@HeikoS | what happens? | 18:24 |
@HeikoS | how is speed affected? | 18:24 |
Saurabh7 | hmm | 18:24 |
Saurabh7 | I need to check , havent run any parallel algo yet | 18:25 |
@HeikoS | ok | 18:25 |
@HeikoS | might be a nice little side thing | 18:25 |
@HeikoS | I think first step is to read up on this case | 18:25 |
@HeikoS | lambday seemed to remember that doing this is actually strongly discouraged | 18:26 |
@HeikoS | so we need to protect ourselves against that | 18:26 |
@HeikoS | apart from that, what algo do you wanna do next? | 18:26 |
Saurabh7 | really, must look into then | 18:26 |
Saurabh7 | HeikoS: I had a look at KNN and CART | 18:26 |
Saurabh7 | CART is really huge, | 18:27 |
@HeikoS | KNN has something in it already hasnt it= | 18:27 |
@HeikoS | I know | 18:27 |
arianepaola | HeikoS would it be possible to add the content from the wiki, containing the description to http://readme.md/? | 18:28 |
@HeikoS | we will do both anyways | 18:28 |
@HeikoS | arianepaola: yes, actually the readme.md should contain the description from the wiki | 18:28 |
@HeikoS | I did that a while ago but it got reverted by an accident I think | 18:28 |
@HeikoS | feel free to send a patch | 18:28 |
arianepaola | ok | 18:28 |
@HeikoS | arianepaola: how is the setup.py going? | 18:29 |
@HeikoS | almost done? | 18:29 |
arianepaola | yes | 18:30 |
arianepaola | wiking will be merging it | 18:30 |
@HeikoS | arianepaola: cool | 18:32 |
@HeikoS | arianepaola: whats the next steps from there? | 18:32 |
arianepaola | the debian package | 18:32 |
@HeikoS | arianepaola: btw, also ping me here if you updated the cookbooks or need help | 18:32 |
@HeikoS | arianepaola: ah yes cool.. | 18:32 |
@HeikoS | arianepaola: and I guess some testing of the setup.py on our buildbot? | 18:33 |
arianepaola | yes | 18:33 |
@HeikoS | arianepaola: ok, looking good | 18:34 |
@HeikoS | keep on pushing it :) | 18:34 |
arianepaola | :-) | 18:35 |
Saurabh7 | HeikoS: then I will be profiling CART now, since its very cluttered , I can then find something to begin with | 18:36 |
@HeikoS | Saurabh7: ok | 18:36 |
@HeikoS | cool | 18:36 |
@HeikoS | Saurabh7: yeah profile it, and benchmark it | 18:36 |
@HeikoS | use datasets with different N, D, and number of relevant features | 18:36 |
@HeikoS | a bit like lasso, but the irrelevant features just contain noise | 18:37 |
@HeikoS | and not sparse | 18:37 |
Saurabh7 | ok cool | 18:38 |
@HeikoS | OXPHOS: hoi | 18:46 |
@HeikoS | OXPHOS: can you give me a brief update ? | 18:47 |
-!- GandalfTheWizard [~Eva@112.10.170.90] has quit [Quit: Leaving.] | 19:04 | |
OXPHOS | HeikoS: sure. I refactored (again) the Linalg Vector class by merging CPU/GPU vector into one vector class and added memory align | 19:35 |
OXPHOS | HeikoS: also started two cookbooks - sharedboost and relaxed tree | 19:35 |
@HeikoS | OXPHOS: cool | 19:36 |
@HeikoS | was mostly after the linalg | 19:36 |
OXPHOS | yep | 19:36 |
@HeikoS | OXPHOS: and next steps for the linalgVector are? | 19:37 |
OXPHOS | transfer data from GPU back to CPU - add test case - (maybe merge Vector to SGVector) - modify linalg class to be compatible with new Vector class | 19:39 |
OXPHOS | HeikoS ^ | 19:39 |
@HeikoS | ok coo | 19:39 |
@HeikoS | I discussed a bit with lambday, I think SGVector should not be merged with linalgVector | 19:40 |
@HeikoS | but wiking is of different opinion, I dont know how strong though | 19:40 |
@HeikoS | but I guess transferring is first | 19:40 |
@HeikoS | cool | 19:40 |
-!- arianepaola [~ariane@unaffiliated/arianepaola] has left #shogun [] | 19:44 | |
-!- arianepaola [~ariane@unaffiliated/arianepaola] has joined #shogun | 19:48 | |
-!- lambday [8028b10a@gateway/web/freenode/ip.128.40.177.10] has quit [Ping timeout: 250 seconds] | 20:41 | |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 21:23 | |
-!- HeikoS [~heiko@nat-204-128.internal.eduroam.ucl.ac.uk] has quit [Quit: Leaving.] | 21:50 | |
arianepaola | bye everyone, see you all tomorrow | 22:04 |
-!- sonney2k_ [~shogun@7nn.de] has quit [Ping timeout: 276 seconds] | 23:09 | |
-!- shogun-toolbox [~shogun@7nn.de] has quit [Ping timeout: 276 seconds] | 23:09 | |
--- Log closed Tue Jun 14 23:09:40 2016 | ||
--- Log opened Tue Jun 14 23:09:47 2016 | ||
-!- shogun-toolbox [~shogun@7nn.de] has joined #shogun | 23:09 | |
-!- Irssi: #shogun: Total of 12 nicks [2 ops, 0 halfops, 0 voices, 10 normal] | 23:09 | |
-!- Irssi: Join to #shogun was synced in 6 secs | 23:09 | |
-!- sonney2k [~shogun@7nn.de] has joined #shogun | 23:09 | |
--- Log closed Wed Jun 15 00:00:40 2016 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!