--- Log opened Tue Apr 12 00:00:36 2011 | ||
-!- ameerkat [~ameerkat@184-98-140-155.phnx.qwest.net] has joined #shogun | 00:59 | |
Gagan | sonney2k: oh damn, you're sleeping I take it | 03:20 |
---|---|---|
-!- Gagan [~brocolli@nat-fw.cs.ualberta.ca] has quit [Quit: Bye!] | 03:20 | |
-!- alesis-novik [~alesis@188.74.87.84] has joined #shogun | 03:58 | |
serialhex | brocolli: whats up? how ya been??? | 06:19 |
serialhex | afk-ing i take it... *goes and washes dishes while shogun compiles* | 06:22 |
-!- akhil_ [75d35896@gateway/web/freenode/ip.117.211.88.150] has joined #shogun | 07:30 | |
-!- dvevre [b49531e3@gateway/web/freenode/ip.180.149.49.227] has quit [Ping timeout: 252 seconds] | 07:55 | |
serialhex | sonney2k: are you here? | 08:31 |
serialhex | i guess not... oh well, if you notice this message i will probably be on for a little while longer... it's ~9am there and it's ~3am here, so i'm looking to go to sleep soon. though i'll check back before i do | 08:46 |
@sonney2k | serialhex, don't work too long hours all the time... anyways I am back | 09:00 |
serialhex | sonney2k: i'll try not to, did you get my latest message on the normal distribution thing? | 09:01 |
@sonney2k | just reading it | 09:01 |
serialhex | ok... do you want me to keep it that way or change it so there are no default args?? | 09:02 |
serialhex | that was my only real concern at this point | 09:02 |
@sonney2k | serialhex, I am just wondering if it would be better to use a templated function for that and then introduce two flavours sth. like rand_float and rand_double | 09:04 |
serialhex | umm... ok, i'll have to find out about templated functions, but it's something new for me to lear right? :D | 09:05 |
@bettyboo | yeah :> | 09:05 |
serialhex | and just so you know, these past few days of intense coding has been really benificial to me, even if i'm not accepted this has been really cool! | 09:06 |
* serialhex had C++ book out & reading about template functions now | 09:07 | |
-!- josip [~josip@unaffiliated/josip] has quit [Remote host closed the connection] | 09:08 | |
serialhex | ...so if i understand it right, i have the templated function that does all the heavy lifting, and then rand_float & rand_double that call the templated function with those types of values... am i right | 09:14 |
serialhex | ? | 09:14 |
@sonney2k | serialhex, I just realize that templates won't be any good if you have to return integers for example | 09:16 |
@sonney2k | ahh just do it withouth default arguments then and I will apply your patch | 09:17 |
serialhex | ok... you do realize that i only have floating-point valued functions there right? | 09:17 |
@sonney2k | serialhex, yes but for templated functions it would also do sth. for int etc | 09:17 |
serialhex | ok, cool... editing & commiting now | 09:18 |
@sonney2k | that would then be nonsense | 09:18 |
serialhex | :D YAY nonsense! | 09:19 |
serialhex | you should see it now | 09:19 |
@sonney2k | I mean if you add a convenience function randn that just calls your normal_random() function but with mean 0 and std-dev 1 then we have all we wanted. | 09:20 |
serialhex | and again, thank you for this opportunity to do all this. this 'pre GSoC competition' has really taught me alot | 09:21 |
serialhex | ...? and why didnt i think of that | 09:21 |
* serialhex goes and edits some more :P | 09:21 | |
@sonney2k | serialhex, don't forget to change the comment too :) | 09:21 |
@bettyboo | yeah ;> | 09:21 |
serialhex | oh, what type would you want it to me? float32_t | 09:21 |
@sonney2k | both | 09:21 |
serialhex | so like randn_32 & randn_64 or something?? | 09:22 |
-!- siddharth [~siddharth@117.211.88.150] has quit [Read error: Connection reset by peer] | 09:22 | |
@sonney2k | float32_t and 64 ... but then name it randn_float / randn_double | 09:22 |
serialhex | actually.. i know hd on | 09:22 |
serialhex | ahh, ok | 09:22 |
serialhex | sonney2k, how is this for comment for each of those fumctions: | 09:29 |
serialhex | // Convenience method for generating Standard Normal random numbers | 09:29 |
serialhex | // Double: Mean = 0 and Variance = 1 | 09:29 |
serialhex | err... 3 /'s for each of those | 09:30 |
serialhex | idk why it only put 2 | 09:30 |
-!- yayo3 [~jakub@ip-78-45-113-245.net.upcbroadband.cz] has joined #shogun | 09:35 | |
serialhex | sonney2k: i just noticed the comment for the abs() function is: ///return the maximum of two integers -- shouldnt it be ///returns the absolute value of a number ?? | 09:37 |
@sonney2k | serialhex, yes | 09:37 |
@sonney2k | copy paste bug... | 09:37 |
@sonney2k | yes good | 09:38 |
serialhex | kinda figured... i'll change that too im my commit | 09:38 |
@sonney2k | Heh | 09:42 |
serialhex | sonney2k: you should see it now | 09:42 |
@sonney2k | serialhex, I am happy with it. will apply it now. | 09:43 |
serialhex | ok cool! i'm going to sleep now. have a good day at work! | 09:44 |
CIA-31 | shogun: Justin Patera master * r74c94a5 / src/libshogun/lib/Mathematics.h : | 09:44 |
CIA-31 | shogun: Changed mornal_random so that only the float32_t version has default arguments. | 09:44 |
CIA-31 | shogun: Unfortunately having float32_t foo(float32_t=1.0) && float64_t | 09:44 |
CIA-31 | shogun: foo(float64_t=1.0) is ambiguous in SWIG, as it does not follow the typedefs or | 09:44 |
CIA-31 | shogun: namespace qualifiers. See "6.15.3 Ambiguity resolution and renaming" and the | 09:44 |
CIA-31 | shogun: "Notes on %rename and %ignore" in the SWIG documentation for more information. - | 09:44 |
CIA-31 | shogun: http://bit.ly/hIB1oY | 09:44 |
CIA-31 | shogun: Soeren Sonnenburg master * rbda6e6e / src/libshogun/lib/Mathematics.h : Merge branch 'master' of https://github.com/serialhex/shogun - http://bit.ly/eHWOA1 | 09:44 |
CIA-31 | shogun: Justin Patera master * r19c35fd / src/libshogun/lib/Mathematics.h : Killing the default arguments on normal_random. - http://bit.ly/fBy2qy | 09:44 |
CIA-31 | shogun: Justin Patera master * rb677245 / src/libshogun/lib/Mathematics.h : Made convenience functions randn_float & randn_double for standard normal random variables. Also edited comment on abs() function to reflect what it truly does. - http://bit.ly/hhhelI | 09:44 |
-!- CIA-31 was kicked from #shogun by bettyboo [flood] | 09:44 | |
-!- CIA-8 [~CIA@208.69.182.149] has joined #shogun | 09:44 | |
serialhex | YAY!!! i feel spechul!! :P | 09:44 |
-!- blackburn [~qdrgsm@188.168.4.201] has joined #shogun | 09:48 | |
yayo3 | god damn macos | 09:49 |
@sonney2k | yayo3, burned your fingers? | 09:49 |
yayo3 | sonney2k: yep. well, not realy. | 09:49 |
yayo3 | I trying to do something with the configure bug. which is actually mac os gcc design decision | 09:50 |
@sonney2k | I see | 09:50 |
yayo3 | aparrently the only way to make it not to generate debug symbols in external files is to compile and link in two steps | 09:50 |
yayo3 | because otherwise the gcc gets too clever | 09:50 |
@sonney2k | yayo3, isn't it enough to drop the -g flag? | 09:51 |
yayo3 | that should work, unless the configre needs it | 09:52 |
yayo3 | configure* | 09:52 |
@sonney2k | yayo3, well we compile with -g on to have some info when crashes happen | 09:53 |
yayo3 | this is actually just the problem with the configure | 09:56 |
yayo3 | it tests the compiler and on macos, you get directories with external debug symbols | 09:57 |
yayo3 | which the remains, as the confire don't know they're there and don't clean them up | 09:57 |
-!- siddharth [~siddharth@117.211.88.150] has joined #shogun | 09:57 | |
yayo3 | also, make clean does not expect the configure-SOMETHING temp files to be directories and crashes | 09:58 |
@sonney2k | yayo3, why don't you add a line that does rm -rf configure-* ? | 09:59 |
yayo3 | sonney2k: well, we takled about that yesterday (or day before yesterday?) and you said rm -rf is a bad idea :) | 10:00 |
@bettyboo | funny | 10:00 |
@sonney2k | yayo3, I thought you do intended to replace rm -f with rm -rf everywhere in make clean | 10:00 |
yayo3 | anyway according to mailing list, rm -rf is what the makers of autoconf decided to do :) | 10:01 |
@sonney2k | yayo3, adding a line with specific rm -rf is fine to me | 10:01 |
yayo3 | cool! I'll do that and send a pull request | 10:01 |
@sonney2k | thanks | 10:07 |
-!- siddharth [~siddharth@117.211.88.150] has quit [Quit: Leaving] | 10:11 | |
-!- siddharth [~siddharth@117.211.88.150] has joined #shogun | 10:12 | |
-!- ameerkat [~ameerkat@184-98-140-155.phnx.qwest.net] has quit [Ping timeout: 260 seconds] | 10:18 | |
-!- akhil_ [75d35896@gateway/web/freenode/ip.117.211.88.150] has quit [Quit: Page closed] | 10:26 | |
-!- blackburn [~qdrgsm@188.168.4.201] has quit [Ping timeout: 240 seconds] | 10:33 | |
-!- yayo3 [~jakub@ip-78-45-113-245.net.upcbroadband.cz] has quit [Quit: Lost terminal] | 10:58 | |
-!- skydiver [c255a037@gateway/web/freenode/ip.194.85.160.55] has joined #shogun | 12:08 | |
-!- siddharth [~siddharth@117.211.88.150] has quit [Remote host closed the connection] | 12:36 | |
-!- siddharth [~siddharth@117.211.88.150] has joined #shogun | 12:43 | |
-!- skydiver [c255a037@gateway/web/freenode/ip.194.85.160.55] has quit [Ping timeout: 252 seconds] | 13:03 | |
-!- josip [~josip@unaffiliated/josip] has joined #shogun | 13:32 | |
josip | sonney2k: should we (gsoc applicants) expect some kind of tasks until the decisions are made?? | 13:54 |
@sonney2k | josip, it is as easy as this: the more you do that separates you from the other students the more you are likely to be in of course. this does not mean we can guarantee your acceptance though. not enough mentors (we can only handle 10 students I think) and fewer slots can be a problem | 13:56 |
josip | hm, okay. So it is note that we will get tasks directly from you, but we have to choose some issues and fix them | 13:57 |
@sonney2k | josip, in any case patches in any direction are most welcome. | 13:57 |
josip | yep, I thought that we'd get specific tasks that are directly related to the proposals | 13:57 |
@sonney2k | josip, ask your potential mentor then.. sometimes it is really hard to suggest topics that are easy enough to be finished in a few hours | 13:59 |
-!- yayo3 [~jakub@195.113.242.138] has joined #shogun | 13:59 | |
@sonney2k | josip, I mean we don't mind starting you to work on the topic you've chosen - but sometimes you have to do a bug chunck of work before it becomes useful at all | 13:59 |
josip | hm, okay - thanks ! | 14:00 |
@sonney2k | josip, what could be done currently is adding examples to all the new kernels in different languages | 14:00 |
@sonney2k | that is easy - no idea if this is something that you would be motivated to do though | 14:00 |
josip | I will also take a look at the issues tracker. Problem is - I'll be travelling a lot this week, hopefully I'll find time to do something | 14:01 |
@sonney2k | josip, then I suggest to do small things | 14:01 |
@sonney2k | and contact your potential mentor ... | 14:02 |
josip | I already did before applying | 14:02 |
yayo3 | so maybe this is old news, but current master branch doesn't compile | 14:02 |
@sonney2k | josip, well then again | 14:02 |
@sonney2k | yayo3, really? | 14:02 |
* sonney2k checks | 14:02 | |
yayo3 | looks that way | 14:03 |
@sonney2k | yayo3, indeed | 14:04 |
@sonney2k | my oversight | 14:04 |
CIA-8 | shogun: Soeren Sonnenburg master * r248812b / src/libshogun/lib/Mathematics.h : fix minor compile error - hurray copy & paste - http://bit.ly/i1Qx9d | 14:04 |
@sonney2k | yayo3, did you say it does not compile? well it does ;-) | 14:05 |
@bettyboo | sonney2k, ;> | 14:05 |
yayo3 | thanks :) | 14:05 |
yayo3 | it gets' kind of easier to play with compiler options when the code compiles :) | 14:05 |
@bettyboo | ;D | 14:05 |
@sonney2k | yayo3, thanks for the report! | 14:08 |
* sonney2k thinks that some kind of automated build testing would be cool | 14:08 | |
@sonney2k | anyone aware of a free service for that? | 14:08 |
yayo3 | opensuse build service can autobuild packages | 14:13 |
yayo3 | no idea whether it does autotesting or somesuch | 14:13 |
@sonney2k | yayo3, packages is not really what is required here - just after each commit a rebuild trigger... | 14:16 |
@sonney2k | dammed, another build problem | 14:16 |
* sonney2k notes to check compilation more carefully, compiling for more interfaces | 14:17 | |
siddharth | sonney2k, yes that would be great...we have to build each time | 14:17 |
@sonney2k | I suggest to use ccache | 14:17 |
@sonney2k | this can turn a full shogun build to last only 5 seconds ;) | 14:18 |
@bettyboo | ho ;> | 14:18 |
* yayo3 is installing ccache NOW | 14:19 | |
siddharth | sonney2k, I went through the code of SGD-Qn and there was one function printQnInfo.It s been defined but never called.so should I include this? | 14:20 |
@sonney2k | siddharth, probably not | 14:26 |
siddharth | ok | 14:26 |
@sonney2k | siddharth, did you get it to compile at least? | 14:27 |
@sonney2k | I mean with SGD-QN in shogun already? | 14:27 |
* sonney2k hopes that this question cannot be misunderstood | 14:28 | |
siddharth | sonney2k, I mean should I include the function when integrating SGD-QN in shogun | 14:29 |
@sonney2k | ^^ that was what my probably not was referring to - no | 14:30 |
@sonney2k | siddharth, I usually start merging algorithms by copying their most important class there and then derive it from e.g. CSVM etc and then see how it goes... | 14:33 |
siddharth | sonney2k, yes I was also looking at the imp classes and functions...but I found this function lying around :P | 14:35 |
siddharth | sonney2k, thanks for the advice :) | 14:38 |
yayo3 | is it just me, or is the configure kinda messy? :) | 14:41 |
@bettyboo | ;> | 14:41 |
-!- skydiver [4deac315@gateway/web/freenode/ip.77.234.195.21] has joined #shogun | 14:44 | |
@sonney2k | yayo3, it is | 14:46 |
@sonney2k | stemming from mplayer's config shell script | 14:46 |
@sonney2k | and grown up to this mess (tm) | 14:47 |
yayo3 | yep. well we now have a working --enable-debug and --disable-debug | 14:47 |
yayo3 | instead of the options being there, saying that debug is defaultly disabled and doing nothing :) | 14:47 |
yayo3 | pull request in a minute :) | 14:47 |
@bettyboo | yeah ;D | 14:47 |
yayo3 | or after I make configure delete configure-BLA.dSYM files after itself anyway | 14:48 |
@sonney2k | mlsec, recently betty is smiling a bit too often | 14:48 |
@bettyboo | sonney2k: well done betty! | 14:48 |
* sonney2k hopes that this doesn't trigger an endless conversion | 14:48 | |
@sonney2k | ok | 14:49 |
@sonney2k | yayo3, did you send the pull already? | 15:05 |
yayo3 | no, I'm still wokring on the "or after I make configure clean up after itself anyway" | 15:08 |
yayo3 | but I can send what I have | 15:08 |
yayo3 | which I just did | 15:10 |
yayo3 | ok I added a cleanup of the .dSYM directories at the end of configure | 15:18 |
yayo3 | (and pulled and stuff) | 15:18 |
@sonney2k | yayo3, your patch goes a bit beyond... | 15:21 |
@sonney2k | if I apply it we need another option for non-optimized compilation (only with -O0) | 15:22 |
yayo3 | oh, you wanted just the -g back? | 15:24 |
yayo3 | well how does --disable-optimization option that disables optimization and cpu detection sounds? :) | 15:25 |
yayo3 | or should the --disable-cpudetection disable also the -O9 -fsomething stuff? | 15:27 |
@sonney2k | yayo3, changing one knob's meaning will generally kill lots of other things | 15:34 |
@sonney2k | IIRC disabe-cpudetection will just use -O2 | 15:35 |
yayo3 | I have added a new option. take a look at it please | 15:49 |
alesis-novik | Good morning (?) | 15:49 |
alesis-novik | I was wondering on what the other students are working on, so that I wouldn't do the same thing | 15:50 |
@sonney2k | alesis-novik, I guess here is the right place to ask them :) | 15:54 |
@sonney2k | not sure what you intend to do | 15:54 |
alesis-novik | I intend to try implementing MLP with a single hidden layer | 15:55 |
CIA-8 | shogun: Soeren Sonnenburg master * r1e6858d / (5 files in 3 dirs): Merge branch 'master' of https://github.com/yayo3/shogun (+6 more commits...) - http://bit.ly/ehcXIW | 15:57 |
@sonney2k | alesis-novik, I am pretty sure no one is working on this - but there is ample toolboxes around that do MLP's (torch, shark,...) so it is potentially only of academic value | 15:57 |
@sonney2k | yayo3, finally! | 15:58 |
yayo3 | yay! :) | 15:58 |
@sonney2k | thanks for your patience with this configure script mess(tm) | 16:00 |
alesis-novik | But then again other toolboxes have half of the stuff shogun has anyway | 16:01 |
yayo3 | anyway I need to run. see you all later | 16:02 |
@sonney2k | alesis-novik, I am just saying that integrating other state-of-the-art MLP implementations would probably be better. That said it does not hurt to have an MLP in shogun as a reference | 16:02 |
@sonney2k | yayo3, l8r! | 16:02 |
-!- yayo3 [~jakub@195.113.242.138] has quit [Quit: leaving] | 16:02 | |
alesis-novik | What other things would you suggest working on? | 16:03 |
@sonney2k | alesis-novik, all ideas I had are in this ideas list | 16:11 |
@sonney2k | if you can implement an MLP rather quickly (1-2 days) do it - otherwise look at other big ml toolboxes (like Weka or shark) and check whether you find anything interesting that is still missing | 16:12 |
alesis-novik | sonney2k, Kernel PCA doesn't look too difficult to implement. | 16:23 |
@sonney2k | alesis-novik, it is partially there in preproc/KPCACut | 16:24 |
@sonney2k | but it needs features like extracting only the top n components and it can currently not be applied | 16:24 |
@sonney2k | just the code for computing is there | 16:24 |
@sonney2k | same with PCACut (just PCA) - while it can be applied it is missing features to cut out the top n components - that would be a nice addition | 16:25 |
@sonney2k | (so far one can only say one wants the all components with eigenvalues > epsilon) | 16:26 |
alesis-novik | sonney2k, I guess I could do a quick patch for that | 16:27 |
@sonney2k | alesis-novik, that would be nice... the other things that need some work can be found by searching for SG_UNSTABLE or SG_NOTIMPLEMENTED | 16:32 |
alesis-novik | thanks for the tip | 16:35 |
-!- siddharth [~siddharth@117.211.88.150] has quit [Ping timeout: 258 seconds] | 16:44 | |
@mlsec | Ok. | 17:04 |
-!- siddharth [~siddharth@117.211.88.150] has joined #shogun | 17:08 | |
brocolli | so, sonney2k, started sifting through the applicants? ;) | 17:13 |
@sonney2k | brocolli, more or less through I would say | 17:14 |
@sonney2k | today google will assign initial slots | 17:14 |
brocolli | i see | 17:14 |
@sonney2k | or better tomorror in germany | 17:14 |
brocolli | have you prioritized the candidates | 17:14 |
@sonney2k | yes | 17:14 |
brocolli | i see | 17:14 |
@sonney2k | the top twenty sth | 17:14 |
brocolli | if we weren't notified, we're not on the upper parts of the list, I take it? | 17:14 |
@sonney2k | notification is only next week IIRC | 17:15 |
brocolli | oh i see | 17:15 |
brocolli | can we find out our # if we ask? in private of course, and understanding that we won't reveal this to anyone ? | 17:15 |
@sonney2k | 25th | 17:15 |
brocolli | noooooo | 17:16 |
brocolli | i didn't think I was that low :( | 17:16 |
brocolli | lol | 17:16 |
@bettyboo | yeah! | 17:16 |
@sonney2k | 25th of april! | 17:16 |
brocolli | oiuc | 17:17 |
brocolli | oic* | 17:17 |
@sonney2k | is the big day with the announcements | 17:17 |
@sonney2k | before that we were asked not to say a word | 17:17 |
brocolli | oh i see. | 17:17 |
brocolli | So even your own list isn't allowede to be discussed because essentially the top n (if n is the number of allotted slots) get them, if they want it | 17:18 |
@sonney2k | brocolli, does that mean your proposal got not accepted and you are now availabe for GSoC? | 17:19 |
@sonney2k | brocolli, btw, I don't even know your real name... | 17:19 |
brocolli | ;) | 17:19 |
brocolli | I guess my institution would give me away | 17:19 |
@bettyboo | he ;D | 17:19 |
brocolli | University of Alberta | 17:19 |
@sonney2k | brocolli, not the interface I have | 17:20 |
@sonney2k | your melange id or so | 17:20 |
brocolli | whats my melange id? :/ | 17:20 |
brocolli | oh my id | 17:21 |
brocolli | its broly | 17:21 |
josip | Can shogun display ROC curves of an SVM ? | 17:21 |
@sonney2k | josip, only compute :) | 17:22 |
@sonney2k | look at performancemeasure | 17:22 |
@sonney2k | in evaluation/ dir | 17:22 |
josip | thanks | 17:23 |
@sonney2k | josip, but the implementation is not really ideal I would say - in particular for precision recall curves. that would be some (relatively easy) task to do | 17:23 |
josip | why not ideal? | 17:23 |
@sonney2k | brocolli, msg me in private... | 17:23 |
@sonney2k | josip, look at the code and you will know - slow slwo slow | 17:23 |
josip | oh, okay | 17:23 |
josip | it retrains for every parameters? :D | 17:23 |
@bettyboo | ;D | 17:23 |
josip | parameter* | 17:23 |
@sonney2k | roc is ok though - but yeah I think this whole class needs polishing | 17:24 |
@sonney2k | I have a python script lying around that does all that - I would love to have that all translated ftp://largescale.ml.tu-berlin.de/largescale/scripts/evaluation.py | 17:26 |
josip | translated to ? | 17:26 |
josip | this is some evalation script for the large scale challenge right | 17:29 |
@sonney2k | (that I wrote yes) | 17:29 |
@sonney2k | it contains many evaluation algorithms | 17:30 |
@sonney2k | that area mostly already in shogun - but some are not | 17:30 |
josip | ok | 17:31 |
@sonney2k | s/area/are/ | 17:32 |
-!- josip [~josip@unaffiliated/josip] has quit [Remote host closed the connection] | 17:41 | |
-!- skydiver [4deac315@gateway/web/freenode/ip.77.234.195.21] has quit [Quit: Page closed] | 17:44 | |
brocolli | so the expected hours are 40 a week, and you get a total of 6k over the 4 months? | 17:55 |
-!- blackburn [~qdrgsm@188.168.2.135] has joined #shogun | 18:07 | |
blackburn | hi all | 18:08 |
serialhex | whats up blackburn?? | 18:11 |
blackburn | it's ok, how are you? :) | 18:12 |
@bettyboo | yeah ;D | 18:12 |
serialhex | chilling, working on getting my repository up-to-date | 18:13 |
blackburn | http://www.google-melange.com/gsoc/events/google/gsoc2011 today is start of 'mentor slots are announced..' | 18:14 |
serialhex | what about you? | 18:14 |
serialhex | cool cool | 18:15 |
blackburn | just came home from my damn university :) | 18:15 |
@bettyboo | yep blackburn | 18:15 |
* serialhex hopes there are a BUNCH of slots for shogun | 18:15 | |
blackburn | 20? :) | 18:15 |
@bettyboo | funny. | 18:15 |
serialhex | nice, it's like dinnertime for you right? | 18:15 |
blackburn | 20-15 | 18:16 |
serialhex | yeah, 20 would be nice :P but it kinda dosnt matter if i get it | 18:16 |
@bettyboo | yep ;D | 18:16 |
blackburn | something became calm, only 11 people in #shogun and only a few commits today | 18:17 |
serialhex | yeah, i guess people are running out of small ideas to commit | 18:18 |
blackburn | there are no kernels anymore :) | 18:19 |
siddharth | lol yeah | 18:19 |
@bettyboo | rahaha | 18:19 |
serialhex | yeah, i noticed they were getting eaten up | 18:20 |
siddharth | graph kernels left :P | 18:20 |
@bettyboo | hihi ;D | 18:20 |
siddharth | bettyboo u dont do anything and just laugh at us | 18:21 |
@bettyboo | siddharth: it would really make sense to do the octave / and older matlab binary formats too... for that look also at the source of octave (it's linked to at the gsoc page) | 18:22 |
serialhex | and then she just talks nonsense to us!! :P | 18:22 |
blackburn | bettyboo: make doc | 18:25 |
@bettyboo | blackburn: well i did make install and now I can use the kernel function which I made | 18:25 |
blackburn | bettyboo: make doc not kernels | 18:26 |
@bettyboo | blackburn: vodka? beer? make doc! | 18:26 |
serialhex | lol! that rocked!!! | 18:28 |
blackburn | leaving now, going to study something :) | 18:29 |
blackburn | see you | 18:29 |
@bettyboo | ;D | 18:29 |
-!- blackburn [~qdrgsm@188.168.2.135] has quit [Quit: Leaving.] | 18:29 | |
-!- brocolli [ylorb@unaffiliated/brocolli] has left #shogun [] | 18:31 | |
@mlsec | well, guess she needs to restart and get less active | 18:32 |
-!- bettyboo [~bettyboo@bane.ml.tu-berlin.de] has quit [Remote host closed the connection] | 18:33 | |
-!- bettyboo [~bettyboo@bane.ml.tu-berlin.de] has joined #shogun | 18:33 | |
-!- mode/#shogun [+o bettyboo] by ChanServ | 18:34 | |
@mlsec | better now | 18:35 |
@mlsec | at least betty has learned a lot about shogun! | 18:36 |
@bettyboo | mlsec: so google will not give us all the slots | 18:36 |
alesis-novik | bettyboo, way to ruin the mood | 18:38 |
@bettyboo | alesis-novik: oh i did that....i think they were unwell | 18:38 |
@mlsec | let's see how many slots we get | 18:38 |
@mlsec | shogun seems to have exceptionally many applications. | 18:39 |
CIA-8 | shogun: Jakub Jirku master * r36bbddd / (2 files): Documentation formula fix - http://bit.ly/eOgq8k | 19:27 |
CIA-8 | shogun: Soeren Sonnenburg master * r8b483bf / (2 files): Merge branch 'master' of https://github.com/yayo3/shogun - http://bit.ly/hmOS7j | 19:27 |
@mlsec | Maybe we should freeze devel on pcacut | 19:44 |
@mlsec | Until gsoc starts and the dim.reduction task is taken care of | 19:45 |
-!- akhil_ [75d35896@gateway/web/freenode/ip.117.211.88.150] has joined #shogun | 20:24 | |
@sonney2k | mlsec, I don't agree :) I think we should start to freeze topics once we know who will occupy them (work on them). | 20:52 |
@bettyboo | lol | 20:52 |
-!- siddharth_ [~siddharth@117.211.88.150] has joined #shogun | 21:10 | |
-!- siddharth [~siddharth@117.211.88.150] has quit [Ping timeout: 258 seconds] | 21:13 | |
-!- yayo3 [~jakub@ip-78-45-113-245.net.upcbroadband.cz] has joined #shogun | 21:21 | |
yayo3 | hey | 21:21 |
@mlsec | okay | 21:25 |
@mlsec | lol | 21:25 |
-!- akhil_ [75d35896@gateway/web/freenode/ip.117.211.88.150] has quit [Quit: Page closed] | 21:29 | |
serialhex | some interesting articles on building a scientific cluster with only consumer hardware & OSS - part 1: http://goo.gl/m7ye3 part 2: http://goo.gl/JIWlq they're going to be using it for neural network simulation apparently, & part 3 will deal with that | 21:31 |
-!- skydiver [c255a037@gateway/web/freenode/ip.194.85.160.55] has joined #shogun | 22:08 | |
@sonney2k | hmmhh I just heard that new orgs rarely get more than 2 slots | 22:15 |
* sonney2k cries | 22:15 | |
* serialhex cries with sonney2k | 22:15 | |
serialhex | that sucks! | 22:15 |
@sonney2k | we won't know before tomorror morning | 22:16 |
yayo3 | what is this I don't even | 22:16 |
serialhex | yeah, thats not cool | 22:16 |
serialhex | but we'll see! | 22:16 |
* yayo3 starts to think it was a bad idea apply only to shogun | 22:17 | |
serialhex | yeah, probably not smart... | 22:18 |
* serialhex hopes he gets this one! | 22:18 | |
-!- blackburn [~qdrgsm@188.168.2.135] has joined #shogun | 22:20 | |
@sonney2k | http://www.booki.cc/gsoc-mentoring/_v/1.0/selecting-the-projects/ says "Every accepted organization is allocated at least one slot. First-year | 22:23 |
@sonney2k | > organizations rarely get more than five slots." - so there is hope. | 22:23 |
serialhex | yes there is!! | 22:24 |
blackburn | oh, talking about slots, I'm interested :) | 22:24 |
blackburn | what I missed? | 22:24 |
yayo3 | blackburn: not much | 22:25 |
yayo3 | just some despair | 22:25 |
yayo3 | and stuff | 22:25 |
blackburn | why despair? | 22:25 |
@sonney2k | blackburn, I heard before that new orgs get only 2 slots - very very few exceptions - but the above statement is at least some hope. I mean we really got 10 excellent proposals. | 22:26 |
serialhex | *woe is all who do not get selected to be part of the summer of code* | 22:26 |
blackburn | heard about 12 | 22:26 |
blackburn | :D | 22:26 |
serialhex | *woe woe, oh woe* | 22:26 |
serialhex | :P | 22:26 |
@sonney2k | serialhex, shall I use your new randn function for that? | 22:26 |
serialhex | HAHA!! yeah, sure... i'll set me as one of the default args :P | 22:27 |
blackburn | sonney2k: think you may discuss it with some googler | 22:27 |
@sonney2k | blackburn, well I did as much as I could already - I won't say more. | 22:28 |
blackburn | sonney2k: so you are already discussing it, as I understand, ok :) | 22:28 |
yayo3 | serialhex: well, it gets pretty frustrating | 22:29 |
serialhex | i know, i'm kind of on-edge too | 22:29 |
serialhex | i quit my non-paying job to code for the summer :D | 22:29 |
yayo3 | you might get to work on something interesting, but instead you spend summer coding PHP eshops to pay you rent | 22:29 |
@sonney2k | lets stay tuned - tomorrow morning we will know for sure | 22:29 |
serialhex | yeah... thats true | 22:29 |
blackburn | sonney2k: we will know slot count? | 22:30 |
@sonney2k | blackburn, well end of business day pacific daylight time | 22:30 |
@sonney2k | but I will probably sleep then | 22:30 |
serialhex | cool, well i'll be back later, gtg get some books on C & liner algebra!! | 22:30 |
-!- skydiver [c255a037@gateway/web/freenode/ip.194.85.160.55] has quit [Ping timeout: 252 seconds] | 22:30 | |
blackburn | sonney2k: I'm tangled with all that timezones | 22:30 |
serialhex | ...that's a ling ways away!!! | 22:30 |
yayo3 | serialhex: linear algebra is black magic! :) | 22:30 |
@bettyboo | ;D yayo3 | 22:30 |
yayo3 | serialhex: stay away from it :) | 22:31 |
@bettyboo | ;D | 22:31 |
serialhex | i've been trying, but it keeps haunting me. | 22:31 |
@sonney2k | so in 9 hrs when I get up I will probably have one particularly important email | 22:31 |
blackburn | ah, okay | 22:31 |
blackburn | in 9 hrs, gmt+3? | 22:31 |
blackburn | phew | 22:31 |
blackburn | gmt+1 | 22:31 |
serialhex | pacific time is gmt -8 | 22:31 |
blackburn | i know | 22:31 |
blackburn | about what timezone sonney2k talking about :) | 22:32 |
serialhex | ok ... you were throwing numbers out there & i didnt know what was going on... | 22:32 |
blackburn | so, important email will be sended in late morning, I see :) | 22:32 |
-!- skydiver [c255a037@gateway/web/freenode/ip.194.85.160.55] has joined #shogun | 22:34 | |
serialhex | ok, i'm out for real this time... later all | 22:34 |
siddharth_ | thats sad...only 2 slots? | 22:34 |
@sonney2k | we will see.... | 22:36 |
siddharth_ | fingers crossed | 22:36 |
blackburn | what will you do in case of 2 slots? :) | 22:36 |
@sonney2k | use serialhex's new random function | 22:36 |
siddharth_ | that will be an even harder job | 22:36 |
siddharth_ | lol | 22:37 |
blackburn | it's a not fair! it's normal! | 22:37 |
@sonney2k | blackburn, yes kidding... | 22:38 |
blackburn | me too | 22:38 |
siddharth_ | lets hope for the best | 22:42 |
@sonney2k | all we can do anyway | 22:44 |
blackburn | tomorrow we will say поехали like gagarin :D | 22:53 |
blackburn | sonney2k: remember what it means? | 22:54 |
@sonney2k | blackburn, no | 22:54 |
blackburn | he :) | 22:54 |
blackburn | strange | 22:55 |
blackburn | it is ~'go on' | 22:55 |
blackburn | *yes I will get to sonney2k while he will not remember all the russian words* :D | 22:56 |
skydiver | poehali!:) | 22:58 |
@sonney2k | heh | 22:58 |
blackburn | I mention it because today is (was) 12 April and 50th anniversary of human space flight :) | 22:59 |
* sonney2k was just watching a movie about that | 23:00 | |
@sonney2k | now bedtime again | 23:00 |
@sonney2k | cu you tomorrow | 23:00 |
@sonney2k | I keep you updated | 23:00 |
blackburn | I even study in university named after S.P. Korolyov, a 'chief' of space flights | 23:00 |
blackburn | :D | 23:00 |
blackburn | will wait for your information :) | 23:01 |
-!- yayo3 [~jakub@ip-78-45-113-245.net.upcbroadband.cz] has quit [Quit: leaving] | 23:04 | |
alesis-novik | Apparently Gagarin was making jokes about sausages and vodka before launch | 23:12 |
blackburn | hm! that is a good idea! will drink some vodka now | 23:13 |
blackburn | :D | 23:13 |
alesis-novik | blackburn, you from the old soviet block? | 23:17 |
blackburn | from the cold russia :) | 23:21 |
alesis-novik | Cool. I'm from Lithuania myself, though I study in the UK now. | 23:26 |
blackburn | ah, that's why you have -s name :) | 23:26 |
blackburn | in baltic states I could be sergeys, right? :) | 23:27 |
alesis-novik | I guess it would be Sergejus with a hard j. | 23:30 |
alesis-novik | It's mostly a Lithuanian thing, not sure about Latvia (everyone speaks Russian there anyway). | 23:30 |
blackburn | alesis-novik: and what about you, do you know some russian words? :) | 23:31 |
alesis-novik | oh, I speak russian, can't type it though | 23:32 |
blackburn | a, ty dazhe razgovarivaesh :) | 23:33 |
blackburn | esche odin govoryaschiy po-russki )) | 23:34 |
blackburn | alesis-novik: think you wiil understand :) | 23:35 |
alesis-novik | moi vse rodstviniki v belarusije i ukraine (I hope this makes sense) | 23:35 |
blackburn | hehe, moya devushka iz belorussii | 23:35 |
alesis-novik | esche odin? Kto esche? | 23:35 |
blackburn | esche skydiver iz pitera | 23:36 |
blackburn | i tam gde-to byl paren iz moskwy | 23:36 |
blackburn | daily lessons of russian. only on #shogun :D | 23:37 |
skydiver | totally 4 | 23:38 |
skydiver | blackburn: it would be good reason for sonney2k to remember russian if we write comments in code in russian | 23:39 |
blackburn | ahahaha | 23:40 |
blackburn | it will be legendary | 23:40 |
blackburn | he will try to kill us | 23:40 |
alesis-novik | All for the greater good | 23:40 |
blackburn | int i=0; // vvedenie peremennoy cikla | 23:41 |
skydiver | ;) | 23:41 |
blackburn | oh, we even could call variables in russian | 23:41 |
skydiver | like variable for gagarin | 23:42 |
blackburn | CClassifier* klassifikator_blizhayschego_soseda | 23:42 |
blackburn | variables named after stalin, hruschev and gagarin | 23:42 |
skydiver | void vodka() | 23:42 |
blackburn | stalin could call destructors of objects | 23:42 |
blackburn | bool inline vodka() { return false; } | 23:43 |
blackburn | and temporary variables andropov and chernenko | 23:43 |
blackburn | with no persistance, will disappear right after being used | 23:43 |
alesis-novik | I wonder... | 23:45 |
alesis-novik | if there is a classification/clustering method with CCP as an abbreviation, the class for that would be CCCP | 23:46 |
blackburn | :D | 23:46 |
alesis-novik | Constructor should call Lenin and destructor should call Gorbachev | 23:46 |
blackburn | CCCP::Lenin() | 23:47 |
blackburn | // not tested, do not call | 23:47 |
blackburn | CCCP::Stalin() | 23:47 |
blackburn | CCCP::~Gorbachev() | 23:47 |
skydiver | that looks like another gsoc proposal | 23:48 |
blackburn | :D | 23:48 |
skydiver | for renaming whole shogun | 23:48 |
blackburn | idea "Introducing CCCP classifier with ability to stalinize random objects" | 23:49 |
skydiver | i think google could issue extra slot for it | 23:49 |
blackburn | going to study f-ing numerical methods :) | 23:51 |
alesis-novik | If the idea would go to the top, I think Sergey Brin would definitely give the extra spot | 23:51 |
blackburn | see you later | 23:52 |
alesis-novik | see you blackburn | 23:52 |
blackburn | *tomorrow, I guess, tomorrow we all be here :) | 23:52 |
blackburn | will be here* (what's up with me?) | 23:52 |
-!- blackburn [~qdrgsm@188.168.2.135] has quit [Quit: Leaving.] | 23:53 | |
--- Log closed Wed Apr 13 00:00:36 2011 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!