--- Log opened Fri Sep 30 00:00:45 2011 | ||
-!- Guest56770 [~sonne@g229061048.adsl.alicedsl.de] has quit [Ping timeout: 256 seconds] | 00:52 | |
-!- blackburn [~blackburn@188.168.5.42] has quit [Quit: Leaving.] | 01:23 | |
CIA-3 | shogun: Sergey Lisitsyn master * r95df0a5 / (10 files): Added support of target dimensionality reducing with negative target dimensionality, improved some issues at NLDR preprocessors - http://git.io/92KVbA | 01:25 |
---|---|---|
CIA-3 | shogun: Soeren Sonnenburg master * r7bb0e51 / (src/configure src/shogun/mathematics/Math.h): add lgammal detection to configure - http://git.io/HPt3Zw | 08:24 |
CIA-3 | shogun: Soeren Sonnenburg master * r1f78185 / src/configure : require pkg-config in configure - http://git.io/gB6LxA | 08:24 |
CIA-3 | shogun: Soeren Sonnenburg master * rb18d7b5 / src/configure : fix type pkg-config2 -> pkg-config - http://git.io/t7VDPQ | 08:26 |
-!- sonne|work1 [~sonnenbu@194.78.35.195] has joined #shogun | 10:23 | |
-!- Netsplit *.net <-> *.split quits: @sonney2k, sonne|work, CIA-3 | 10:30 | |
-!- Netsplit over, joins: @sonney2k, CIA-3 | 10:40 | |
-!- sonne|work1 is now known as sonne|work | 14:25 | |
-!- heiko [~heiko@144.82.98.11] has joined #shogun | 15:33 | |
-!- heiko [~heiko@144.82.98.11] has quit [Ping timeout: 260 seconds] | 16:00 | |
-!- blackburn [~blackburn@188.168.4.53] has joined #shogun | 16:53 | |
-!- ChrisD_ [43dc92a2@gateway/web/freenode/ip.67.220.146.162] has joined #shogun | 20:51 | |
ChrisD_ | hi | 20:51 |
blackburn | hi | 20:58 |
ChrisD_ | has anyone seen this error with shogun 1.0.0 when switching to a guassian kernel? [WARN] Thread creation failed (thread 0 of 14) with error:'Success' | 21:09 |
blackburn | ChrisD_: no, what is your os? | 21:11 |
@sonney2k | ChrisD_, do you have a small example to reproduce? | 21:11 |
ChrisD_ | redhat | 21:12 |
@sonney2k | *small and reproducible* | 21:12 |
@sonney2k | and which $LANG is this | 21:12 |
ChrisD_ | python module | 21:12 |
blackburn | sonney2k: would you mind to place arpack and lapack things into mathematics/libs? | 21:13 |
@sonney2k | blackburn, why not just mathematics/ ? | 21:13 |
blackburn | sonney2k: well no difference | 21:13 |
blackburn | hehe | 21:13 |
@sonney2k | ChrisD_, please give us an example such that we can debug | 21:17 |
CIA-3 | shogun: Soeren Sonnenburg master * rb6d03b7 / (6 files): move inline functions from .h to .cpp - http://git.io/FLiwIA | 21:17 |
ChrisD_ | sure just looking for the exact line where this fails | 21:17 |
@sonney2k | ChrisD_, thx | 21:18 |
@sonney2k | gtg | 21:18 |
@sonney2k | cu | 21:18 |
-!- blackburn [~blackburn@188.168.4.53] has quit [Quit: Leaving.] | 21:27 | |
-!- blackburn [~blackburn@188.168.4.53] has joined #shogun | 21:27 | |
@sonney2k | Re | 21:35 |
CIA-3 | shogun: Sergey Lisitsyn master * rf55f818 / (14 files in 2 dirs): Added single precision LAPACK and ARPACK support, changed preprocessors according to new interface - http://git.io/QZd-HQ | 22:03 |
CIA-3 | shogun: Sergey Lisitsyn master * r236bb14 / (8 files in 3 dirs): Merge branch 'master' of github.com:shogun-toolbox/shogun - http://git.io/PKr5ig | 22:03 |
blackburn | sonney2k: single precision is now supported | 22:03 |
blackburn | single precision NLDR preprocessors to go | 22:03 |
@sonney2k | OK | 22:03 |
blackburn | OK? :) | 22:04 |
@sonney2k | you mean in addition to double precision | 22:04 |
@sonney2k | ? | 22:04 |
blackburn | sonney2k: yes | 22:04 |
blackburn | sonney2k: in single precision we could embed even 20K vectors | 22:04 |
blackburn | *on my machine | 22:05 |
blackburn | *without so much pain | 22:05 |
@sonney2k | I see | 22:05 |
blackburn | sonney2k: the only problem I see - how to template only for float32_t and float64_t? | 22:05 |
blackburn | will implementing it only for float{32,64} work? | 22:06 |
blackburn | ah nevermind, I will think about it | 22:06 |
@sonney2k | yes | 22:07 |
blackburn | code duplicating is pretty bad way | 22:07 |
@sonney2k | ChrisD_, blackburn I found a simple way to trigger the issue: just run regression_svrlight_modular.py with 10 threads | 22:07 |
shogun-buildbot | build #303 of libshogun is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/libshogun/builds/303 blamelist: blackburn91@gmail.com | 22:07 |
@sonney2k | ahh we need heiko | 22:08 |
blackburn | oh | 22:08 |
blackburn | my fault | 22:08 |
ChrisD_ | Thanks, how do i go about printing the line number where it occurs? | 22:08 |
CIA-3 | shogun: Sergey Lisitsyn master * rea5f083 / src/shogun/preprocessor/KernelLocallyLinearEmbedding.cpp : Removed wrong include from KLLE - http://git.io/rJAmzQ | 22:09 |
@sonney2k | obj.io.enable_file_and_line() | 22:10 |
@sonney2k | where obj is some shogun object | 22:10 |
@sonney2k | #$#J@KL$J%!?!!! | 22:11 |
@sonney2k | who thinks windows 7 is *the* evolution | 22:12 |
@sonney2k | one cannot even install it | 22:12 |
@sonney2k | blackburn, wouldn't it be nice to have some function in evaluation for the curve stuff that gives you the bias to achieve a certain fpr / tpr etc? | 23:04 |
@sonney2k | blackburn, could you please also move the inline functions from the evaluation/*.h to the .cpp? | 23:10 |
@sonney2k | blackburn, I think it is better to have just one arpack.h/cpp and then implement everything underneath via templates | 23:23 |
@sonney2k | ChrisD_, I don't understand why it is failing... | 23:33 |
@sonney2k | does not make any sense | 23:33 |
ChrisD_ | where can i find the necessary data files to run the example file? | 23:35 |
ChrisD_ | maybe i can make sure that code produces the same error | 23:35 |
@sonney2k | ChrisD_, just do git submodule update --init from the git root dir | 23:47 |
@sonney2k | shogun/ folder | 23:47 |
@sonney2k | (not shogun/src/shgoun) | 23:47 |
blackburn | sonney2k: still here? | 23:54 |
blackburn | (01:04:37 AM) sonney2k: blackburn, wouldn't it be nice to have some function in evaluation for the curve stuff that gives you the bias to achieve a certain fpr / tpr etc? ---------- didn't understand | 23:56 |
blackburn | (01:04:37 AM) sonney2k: blackburn, wouldn't it be nice to have some function in evaluation for the curve stuff that gives you the bias to achieve a certain fpr / tpr etc? -------- will do | 23:57 |
blackburn | (01:23:01 AM) sonney2k: blackburn, I think it is better to have just one arpack.h/cpp and then implement everything underneath via templates ---------- I can't because both single and double precision superlu headers will fail if described in one file | 23:58 |
--- Log closed Sat Oct 01 00:00:48 2011 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!