--- Log opened Sat Jun 02 00:00:41 2012 | ||
heiko2 | hihi | 00:00 |
---|---|---|
heiko2 | heiko3 | 00:01 |
heiko3 | heiko2: so did you submit? | 00:01 |
heiko3 | nips? | 00:01 |
heiko2 | yes | 00:01 |
heiko2 | its done :) | 00:01 |
heiko2 | lots of work | 00:01 |
heiko3 | congrats | 00:01 |
heiko3 | and exam? | 00:01 |
heiko2 | thanks :) | 00:01 |
heiko2 | exam was nice | 00:01 |
heiko2 | also by arthur btw :) | 00:01 |
-!- heiko1 [~heiko@host86-179-58-219.range86-179.btcentralplus.com] has quit [Ping timeout: 260 seconds] | 00:01 | |
heiko3 | hahah exam by supervisor?? | 00:01 |
heiko2 | yes, funny isnt it? :) | 00:01 |
heiko3 | that should be unfair a little | 00:01 |
heiko2 | oh, my girl just returned home, gotta go | 00:02 |
heiko3 | heh | 00:02 |
heiko2 | no, he gave me no hints ;) | 00:02 |
-!- heiko3 is now known as blackburn | 00:02 | |
-!- heiko2 [~heiko@host86-177-14-77.range86-177.btcentralplus.com] has quit [Ping timeout: 260 seconds] | 00:06 | |
n4nd0 | good night! | 01:11 |
blackburn | nite | 01:12 |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving] | 01:13 | |
-!- blackburn [~blackburn@31.28.59.65] has left #shogun [] | 02:19 | |
CIA-9 | shogun: Soeren Sonnenburg master * r78f9ac6 / src/configure : fix octave detection for octave 3.6.x (Closes issue #383) - http://git.io/aDihsw | 06:44 |
CIA-9 | shogun: Soeren Sonnenburg master * r8a15088 / src/configure : fix ruby detection in configure - http://git.io/DfZJdw | 06:44 |
CIA-9 | shogun: Soeren Sonnenburg master * r7344caa / (2 files in 2 dirs): add empty define for ignore in class list - http://git.io/jFNJBg | 06:44 |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 09:03 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving] | 11:25 | |
-!- blackburn [~blackburn@31.28.59.65] has joined #shogun | 12:44 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 14:11 | |
-!- blackburn [~blackburn@31.28.59.65] has quit [Ping timeout: 244 seconds] | 14:12 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving] | 15:12 | |
-!- blackburn [~blackburn@31.28.59.65] has joined #shogun | 18:59 | |
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has quit [Remote host closed the connection] | 20:17 | |
CIA-9 | shogun: Soeren Sonnenburg master * rb921a7d / (2 files in 2 dirs): add proper copyrights in Director* stuff - http://git.io/_GUY1g | 20:24 |
CIA-9 | shogun: Soeren Sonnenburg master * r80b8013 / (src/shogun/lib/Hash.cpp src/shogun/lib/Hash.h): fix compile warning in md5sum - http://git.io/gbxZXQ | 20:24 |
CIA-9 | shogun: Soeren Sonnenburg master * rf0a2a12 / (5 files in 3 dirs): add empty liblinear regressor - http://git.io/e6BGBA | 20:24 |
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has joined #shogun | 20:24 | |
blackburn | haha | 20:26 |
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has quit [Remote host closed the connection] | 20:27 | |
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has joined #shogun | 20:31 | |
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has quit [Remote host closed the connection] | 20:37 | |
-!- gsomix [~gsomix@85.26.235.175] has joined #shogun | 21:18 | |
gsomix | mechanics is full of pain | 21:19 |
gsomix | depressive lagrangians, suffering hamiltonians and many integrals of fear | 21:21 |
gsomix | they come to me in dreams | 21:24 |
gsomix | and organize an orgy of terror | 21:25 |
blackburn | :D | 21:26 |
blackburn | well I still feel excited about Lagrange's stuff | 21:27 |
gsomix | blackburn, my favorite subject is multi-dimensional oscillators | 21:38 |
-!- KMcQuisten [d8338942@gateway/web/freenode/ip.216.51.137.66] has joined #shogun | 22:15 | |
KMcQuisten | Afternoon Sergey | 22:15 |
blackburn | hey | 22:15 |
blackburn | (its me) | 22:16 |
blackburn | :) | 22:16 |
KMcQuisten | I'd guessed :) | 22:16 |
KMcQuisten | So, any ideas as to what the problem is? | 22:16 |
blackburn | okayy let me try to guess | 22:16 |
blackburn | could you please first say what is classifier you do use? | 22:17 |
blackburn | MKLRegression? | 22:17 |
KMcQuisten | I'm working with the MKLRegression | 22:17 |
KMcQuisten | Yes | 22:17 |
blackburn | let me check the code | 22:17 |
blackburn | okay and the problem appears only with combined features, right? | 22:18 |
KMcQuisten | To the best of my knowledge yes. It worked fine when I gave a single kernel to LibSVR, but I dind't try putting a single kernel into MKLRegression | 22:20 |
blackburn | let me try to come with some reproducing example | 22:20 |
blackburn | KMcQuisten: should it look similar to python_modular/kernel_combined_modular.py? | 22:23 |
-!- gsomix [~gsomix@85.26.235.175] has quit [Quit: Ex-Chat] | 22:26 | |
-!- gsomix [~gsomix@85.26.235.175] has joined #shogun | 22:26 | |
KMcQuisten | Yes, essentially, but using StringWordFeatures and the CommWordStringKernel in the Comined objects. | 22:26 |
blackburn | I changed mkl_binclass_modular.py to regression - works | 22:27 |
blackburn | lets see if stringfeatures change the game | 22:28 |
blackburn | KMcQuisten: can I ask you to modify some code meanwhile? some simple thing to debug | 22:31 |
KMcQuisten | Sure | 22:31 |
blackburn | I just want to know where it wants to stop | 22:31 |
KMcQuisten | Right | 22:31 |
KMcQuisten | Just let me know where I need to modify things | 22:32 |
blackburn | shogun/features/StringFeatures.cpp:264 | 22:32 |
blackburn | replace SG_ERROR with SG_WARNING and try to recompile | 22:32 |
blackburn | ah and %d after "is" word in msg (I forgot that) | 22:33 |
KMcQuisten | Ok, done. Now I'll recompile | 22:34 |
blackburn | thanks | 22:34 |
KMcQuisten | Ok, here's what I got. It looks like we have an index issue: | 22:38 |
KMcQuisten | Spectrum_MKLselect.py:66: RuntimeWarning: [WARN] Requested feature vector with index 50 while total num is 50 testsvm.train(combfeats) Traceback (most recent call last): File "Spectrum_MKLselect.py", line 66, in <module> testsvm.train(combfeats) SystemError: [ERROR] Trying to access string[50] but num_str=50 | 22:38 |
blackburn | hmm | 22:40 |
blackburn | only 50, but no more, right? | 22:40 |
KMcQuisten | Yes. I only put in 50 data points as a toy set. Let me check with the whole dataset | 22:40 |
blackburn | wait wait | 22:40 |
blackburn | ERROR is raised somewhere | 22:41 |
KMcQuisten | Yes. I think it's trying to access index 50 on a vector that's 0 indexed with 50 items in ti | 22:41 |
blackburn | no, that can go for bigger values (it stopped because of the SG_ERROR) | 22:42 |
KMcQuisten | How is that possible? If you have a vector of length 50 its highest index is 49. I get the same kind of error with the full set of 2430 sequences | 22:44 |
blackburn | I mean it can go for 51, etc | 22:44 |
KMcQuisten | SystemError: [ERROR] Trying to access string[2430] but num_str=2430 | 22:45 |
KMcQuisten | RIght. I know i'm not limited to 50 data points | 22:45 |
blackburn | ok let me find this ERROR | 22:45 |
blackburn | I tried to combine string and real features btw | 22:45 |
blackburn | works | 22:45 |
blackburn | KMcQuisten: could you please replace SG_ERROR with SG_WARNING at line 348 too? | 22:48 |
KMcQuisten | sure | 22:51 |
KMcQuisten | I will also try and comine the real and the StringWord | 22:51 |
KMcQuisten | and see if it works for me too | 22:51 |
blackburn | what do you combine? | 22:52 |
blackburn | string and string? | 22:52 |
KMcQuisten | StringWord and StringWord. DNA sequences mapped into the unsigned 16bit integers. In python, it's StringWordFeatures, but in the C++ it's StringFeatures<uint_16> | 22:53 |
KMcQuisten | not StringFeatures<char> | 22:53 |
blackburn | ok | 22:54 |
blackburn | KMcQuisten: okay I managed to reproduce | 22:56 |
blackburn | two StringWord feautres | 22:56 |
KMcQuisten | does it work for you? | 22:59 |
blackburn | no dies on the same error | 22:59 |
KMcQuisten | That makes me a happy cmaper | 22:59 |
KMcQuisten | camper | 22:59 |
KMcQuisten | at least it's reproduceable | 23:00 |
blackburn | yeah I'll try to figure out what is wrong | 23:00 |
KMcQuisten | thanks so much! | 23:00 |
KMcQuisten | I've got plenty of other work and writing to do, so i'll just camp here for a while | 23:01 |
blackburn | ok sure | 23:01 |
blackburn | I'll let you know on any progress | 23:02 |
KMcQuisten | Thank you | 23:02 |
gsomix | good night guys | 23:11 |
blackburn | KMcQuisten: hmm as I thought it stopped at 183 | 23:23 |
blackburn | 92*2 | 23:23 |
-!- gsomix [~gsomix@85.26.235.175] has quit [Ping timeout: 260 seconds] | 23:25 | |
KMcQuisten | What does that indicate? | 23:26 |
blackburn | some code adds number of feature vectors somehow I guess | 23:28 |
blackburn | but it is really strange | 23:28 |
KMcQuisten | What's the significance of the 92 * 2? | 23:28 |
blackburn | number of feature vectors in test example is 92 | 23:28 |
KMcQuisten | Ah | 23:29 |
blackburn | and two string word features | 23:29 |
blackburn | KMcQuisten: good thing I detected a bug :D | 23:30 |
KMcQuisten | Indeed. I changed that error to a warning as you asked before and recompiled, got the following: | 23:31 |
KMcQuisten | Spectrum_MKLselect.py:66: RuntimeWarning: [WARN] Requested feature vector with index 2430 while total num is 2430 | 23:32 |
KMcQuisten | Spectrum_MKLselect.py:66: RuntimeWarning: [WARN] Requested feature vector with index 2431 while total num is 2430 | 23:32 |
KMcQuisten | Spectrum_MKLselect.py:66: RuntimeWarning: [WARN] Requested feature vector with index 2432 while total num is 2430 | 23:32 |
KMcQuisten | Spectrum_MKLselect.py:66: RuntimeWarning: [WARN] Trying to access string[2432] but num_str=2430 | 23:32 |
KMcQuisten | Segmentation fault (core dumped) | 23:33 |
KMcQuisten | Does the Combinedfeats object adress all the features as thogh they are one long list? | 23:33 |
blackburn | something like that but I do not understand what is the difference between real+string and string+string | 23:34 |
CIA-9 | shogun: Sergey Lisitsyn master * r3e1dc47 / src/shogun/regression/svr/SVRLight.cpp : Fixed wrong free in SVRLight - http://git.io/n7oILw | 23:37 |
blackburn | KMcQuisten: okay I'm pretty near I think | 23:57 |
blackburn | svr bug | 23:57 |
--- Log closed Sun Jun 03 00:00:41 2012 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!