--- Log opened Thu Nov 24 00:00:13 2011 | ||
--- Day changed Thu Nov 24 2011 | ||
@sonney2k | false alarm? | 00:00 |
---|---|---|
@sonney2k | I mean how is it possible to have an invalid read of size one in this line | 00:00 |
blackburn | sonney2k: ary/ | 00:01 |
blackburn | ? | 00:01 |
@sonney2k | ary is a ptr | 00:01 |
blackburn | how big is void pointer? | 00:01 |
@sonney2k | 8 bytes | 00:01 |
@sonney2k | or 4 | 00:01 |
@sonney2k | depending on 64/32bit system | 00:01 |
blackburn | sonney2k: then I don't know | 00:02 |
blackburn | sonney2k: don't you get segfault with that script? | 00:02 |
@sonney2k | didn't try | 00:04 |
blackburn | could you please? | 00:05 |
blackburn | I'm currently recompiling and will test again | 00:05 |
blackburn | sonney2k: don't you think testing framework is something of interest for gsoc project? | 00:07 |
@sonney2k | yes agreed - but we need a concrete plan for that | 00:08 |
@sonney2k | and it requires heikos migrations | 00:08 |
blackburn | sonney2k: we have time to think about | 00:08 |
blackburn | wow too late already | 00:13 |
blackburn | sonney2k: yes I got segfault again | 00:17 |
blackburn | what a non-sense | 00:17 |
@sonney2k | the other thing I'd like to see is a cool tutorial with applications | 00:17 |
@sonney2k | blackburn, you should compile with optimizations disabled | 00:17 |
blackburn | sonney2k: already | 00:17 |
@sonney2k | and then the usual stuff | 00:18 |
@sonney2k | decision trees | 00:18 |
blackburn | ensembles | 00:18 |
@sonney2k | boosting etc | 00:18 |
@sonney2k | some new ML algos :) | 00:18 |
@sonney2k | anyway I am tired | 00:19 |
@sonney2k | have a nice sleep | 00:19 |
@sonney2k | cu | 00:19 |
blackburn | have a nice sleep too | 00:19 |
blackburn | see you | 00:19 |
blackburn | wow | 00:21 |
blackburn | Labels with same numpy array leads to already freed memory error | 00:21 |
blackburn | that's crazy a lot | 00:21 |
-!- blackburn [~blackburn@188.168.4.253] has quit [Quit: Leaving.] | 00:28 | |
-!- emrecelikten [~emre@92.44.235.73] has joined #shogun | 05:28 | |
emrecelikten | Hello there! Apparently Gaussian HMMs aren't supported in Shogun. Is there anybody working on it? | 05:29 |
-!- emrecelikten [~emre@92.44.235.73] has left #shogun [] | 06:59 | |
-!- emrecelikten [~emre@92.44.235.73] has joined #shogun | 06:59 | |
sonne|work | emrecelikten: noone is working on that | 09:08 |
sonne|work | so you could be first :D | 09:08 |
-!- blackburn [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has joined #shogun | 14:13 | |
blackburn | sonne|work: forgot to say | 14:13 |
sonne|work | yes? | 14:14 |
blackburn | e.g. Labels([1,2,3]) | 14:14 |
blackburn | will lead to problems | 14:14 |
blackburn | and we should do Labels(numpy.array([1,2,3])) | 14:14 |
sonne|work | so the numpy one works the other not? | 14:14 |
blackburn | yes | 14:14 |
blackburn | I have too weak ideas what is going wrong but I know it is | 14:15 |
blackburn | sonne|work: I managed to run this python example yesterday | 14:15 |
sonne|work | great that you figured taht out | 14:16 |
blackburn | sonne|work: if you have shogun available could you please run this: | 14:17 |
blackburn | x = [1,2,3] | 14:17 |
blackburn | for i in range(10): | 14:17 |
blackburn | l = Labels(x) | 14:17 |
blackburn | if it fails I would get more sure about it | 14:17 |
sonne|work | I recall that we had a similar report some time back but no one could reproduce the problem | 14:17 |
blackburn | I shared the folder with that script to madeleine, no report yet | 14:23 |
blackburn | but I hopt it will go smooth this time | 14:23 |
-!- blackburn_ [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has joined #shogun | 15:41 | |
-!- blackburn [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has quit [Ping timeout: 265 seconds] | 15:43 | |
-!- policecar [~policecar@141.23.79.130] has joined #shogun | 15:44 | |
-!- policecar [~policecar@141.23.79.130] has quit [Quit: policecar] | 15:51 | |
-!- blackburn_ [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has quit [Quit: Page closed] | 16:01 | |
-!- policecar [~policecar@141.23.79.130] has joined #shogun | 16:04 | |
policecar | hey, sorry to bother you(s) again but i can't seem to get PCA to work with sparse matrices... i currently run the code below (python) and can't get rid of | 17:18 |
policecar | assertion features->get_feature_class()==C_SIMPLE failed in file preprocessor/PCA.cpp line 71 | 17:18 |
policecar | from shogun.Features import SparseRealFeatures, SparseIntFeatures | 17:18 |
policecar | from shogun.Preprocessor import PCA | 17:18 |
policecar | feats = SparseRealFeatures( data.astype(float) ) | 17:18 |
policecar | 17:18 | |
policecar | preprocessor = PCA() | 17:18 |
policecar | preprocessor.init( feats ) | 17:18 |
policecar | preprocessor.apply_to_sparse_feature_matrix( feats ) | 17:18 |
policecar | where data is a csc_matrix | 17:18 |
policecar | could you gimme any hints as to what i'm doing wrong? | 17:18 |
sonne|work | it doesn't work with sparse matrices - so you need to convert it to dense features and then apply PCA | 17:21 |
policecar | oh | 17:21 |
policecar | hehe, k, thx | 17:21 |
-!- policecar [~policecar@141.23.79.130] has quit [Quit: policecar] | 17:29 | |
-!- blackburn [~blackburn@188.168.2.226] has joined #shogun | 17:30 | |
blackburn | sonney2k: my simple labels test does not reproduce the error anyhow | 19:38 |
@sonney2k | blackburn, it does | 22:06 |
@sonney2k | I can reproduce it here | 22:06 |
@sonney2k | but it is not reliable | 22:06 |
blackburn | sonney2k: hmm ok | 22:06 |
@sonney2k | the issue occurs from time to time only | 22:06 |
@sonney2k | if I use Labels([0.0]*1000) it happens most of the time | 22:07 |
blackburn | sonney2k: I guess it is something of 'list' type nature | 22:07 |
@sonney2k | but I totally don't understand why it crashes sometimes and sometimes not | 22:07 |
@sonney2k | yes but it is converted to narray | 22:08 |
blackburn | sonney2k: is it? | 22:08 |
@sonney2k | but it crashes even before that(!!!) | 22:08 |
blackburn | ah | 22:08 |
blackburn | well no idea then | 22:08 |
@sonney2k | yes via PyArray_FromAny | 22:08 |
@sonney2k | me neither | 22:08 |
blackburn | no one should use lists to have a conversation with shogun :) | 22:09 |
blackburn | 1st commendment | 22:09 |
@sonney2k | we could simply disable that | 22:10 |
@sonney2k | let me just do that | 22:10 |
blackburn | sure, pretty unsafe | 22:10 |
@sonney2k | I still don't understand how it could crash | 22:11 |
@sonney2k | (I am in train home - network could fail at any time) | 22:12 |
blackburn | wow | 22:12 |
blackburn | you really do shogun all the time :D | 22:12 |
@sonney2k | stack overflow or so? | 22:18 |
@sonney2k | argh need to leave train | 22:19 |
@sonney2k | cu | 22:19 |
blackburn | sonney2k: ok | 22:19 |
blackburn | sonney2k: what stackoverflow? | 22:19 |
@sonney2k | blackburn, I mean I cannot make sense of this valgrind error nor the crash | 23:32 |
@sonney2k | and I don't understand why it still wants to accept lists | 23:33 |
@sonney2k | my only idea now is to explicitly reject lists | 23:33 |
blackburn | why not | 23:34 |
-!- blackburn [~blackburn@188.168.2.226] has quit [Quit: Leaving.] | 23:51 | |
--- Log closed Fri Nov 25 00:00:59 2011 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!