2018-11-12 22:39
ChanServ changed the topic of #mlpack to: "mlpack: a fast, flexible machine learning library :: We don't always respond instantly, but we will respond; please be patient :: Logs at
http://www.mlpack.org/irc/
06:27
<
chopper_inbound4 >
Hey Shikhar Jaiswal (Gitter) , are you online? I need some help regarding #2364.
07:36
<
jenkins-mlpack2 >
Yippee, build fixed!
08:34
ImQ009 has joined #mlpack
16:43
marko has joined #mlpack
16:43
<
marko >
i cannot save LARS model in CPP
16:44
<
marko >
here is the code
16:44
<
marko >
/ Declare GLM wh specific parameters.
17:03
<
zoq >
marko: Don't see any code so far.
17:06
<
marko >
line by line
17:06
<
marko >
mlpack::regression::LARS glm(0.25,0.25);
17:06
<
marko >
can you see now?
17:10
<
marko >
glm.Train(x_train, y_train, false);
17:11
<
marko >
mlpack::data::Save("glm.xml","model", glm, false, mlpack::data::format::xml);
17:14
<
jeffin143[m] >
You can share code
17:14
<
jeffin143[m] >
Using pastebin marko
17:14
<
jeffin143[m] >
Also you can add comments there to quickly make us understand :)
17:14
<
jeffin143[m] >
I am guessing the ui here is not good enough
17:15
<
marko >
i try again
17:15
<
jeffin143[m] >
If you need any help in sharing code let me know
17:15
<
marko >
mlpack::regression::LARS glm(0.25,0.25);
17:15
<
zoq >
marko: What is the error you get?
17:15
<
jeffin143[m] >
Yeah it is good :)
17:16
<
jeffin143[m] >
If you are comfortable we are good
17:16
<
marko >
glm.Train(x_train, y_train, false);
17:16
<
marko >
mlpack::data::Save("glm.xml","model", glm, false, mlpack::data::format::xml);
17:17
<
marko >
that is the whole piece of the code
17:17
<
marko >
and the build is not sucessful
17:17
<
zoq >
marko: Okay, do you have error as well?
17:18
<
marko >
'serialize': is not a member of 'arma::Mat<double>'
17:18
<
zoq >
What version are you using?
17:20
<
marko >
W10; VS2017CE; mlpack is installed via vcpkg
17:24
<
marko >
any help from the other side?
17:27
<
zoq >
marko: just did a quick test on lab.mlpack.org and had no issues
17:28
<
marko >
looking at the doc
17:28
<
zoq >
marko: can you run the code and see if that works?
17:28
<
marko >
but save it
17:28
<
marko >
save the model
17:28
<
marko >
that is what i cannot do
17:29
<
marko >
this line fails
17:29
<
zoq >
like this, works as well
17:30
<
marko >
mlpack::data::Save("glm.xml","model", glm, false, mlpack::data::format::xml);
17:30
<
marko >
I will analyze your example
17:30
<
marko >
and try to understand where is the error
17:30
<
marko >
in my program
17:30
<
rcurtin >
marko: I saw you were using 3.1.1; if you can upgrade, perhaps this might help
17:31
<
rcurtin >
zoq: it is so cool that the notebooks are working now... those gists are awesome :)
17:31
<
marko >
vcpg does not offer to upgrade
17:31
<
marko >
it says its all good
17:31
<
marko >
no newer version
17:31
<
rcurtin >
and we just released 3.3.0
17:31
<
zoq >
marko: did you include: #include <mlpack/core.hpp>
17:32
<
marko >
of course...
17:32
<
rcurtin >
I don't know much about vcpkg though so I can't say why the newer version isn't available. you could always file a bug there? :) or maybe you can get it to work with 3.1.1
17:33
<
zoq >
Does vcpkg ship the executables as well?
17:35
<
zoq >
Looks like they do, maybe you can test the lars executable and see if you can save the model?
17:35
<
marko >
but if I understand
17:35
<
marko >
from the standpoint of syntax
17:36
<
marko >
you see nothing wrong with my code
17:37
<
marko >
btw the error resides
17:37
<
marko >
in this file
17:37
<
marko >
c:\vcpkg\installed\x64-windows\include\boost\serialization\access.hpp
17:39
<
zoq >
Looked good to me, couldn't really run it because I don't have the data, but I guess it's pretty similar to what I used for testing?
17:41
<
marko >
all the best guys
17:41
<
marko >
and many, many thanks for help
17:41
<
marko >
i consider the gist instrucive
17:41
<
marko >
i will try to go on with debugging
17:49
<
marko >
btw I live in Augsburg :D
18:11
<
rcurtin >
oops, cat keyboard
18:23
<
marko >
no problem
18:23
<
zoq >
marko: Augsburg, stay safe
18:24
<
marko >
you ca tell zoq that I have ran the code
18:24
<
marko >
and the code works
18:24
<
marko >
now i need to understand why my code is not working
18:24
<
zoq >
okay, strange
18:24
<
marko >
we will do the best
18:24
<
marko >
pandemic is declining here in Germany
18:24
<
marko >
thats good news
18:25
<
marko >
however, we will see what happens after that
18:25
<
marko >
than it becomes critical
18:25
<
zoq >
maybe you can post the code on pastbin or github gist, it's a little bit hard to read in irc :)
18:25
<
marko >
there is really nothing
18:25
<
marko >
except i do not use beta
18:25
<
marko >
and there is decomposition is false
18:26
<
marko >
so i need to understand where is the error
18:26
<
zoq >
marko: That shouldn't change much.
18:26
<
marko >
because i have the XML with your model
18:26
<
zoq >
okay, but that is good.
18:26
<
marko >
many thanks
18:26
<
marko >
and keep up the good work
18:27
<
marko >
signing out
18:27
<
zoq >
I'm in Berlin, so I'm in a similar situtation.
18:27
<
marko >
i have alreaddy starred some of the Github repositories
18:27
<
marko >
that you have
18:27
<
marko >
impressive work
18:28
<
marko >
to be exact
18:28
<
marko >
im data scientist at MAN btw
18:28
<
marko >
so we might push a little bit more with this CPP thing
18:28
<
zoq >
Great, hopefullt mlpack can be of any help :)
18:28
<
marko >
its good environment for ML
18:29
<
marko >
maybe better than Java
18:29
<
marko >
we will se
18:29
<
zoq >
Right, I guess python is strong too.
18:29
<
marko >
there is slightly more emphasis on classification tasks
18:29
<
marko >
I would say that regression tasks are underrepresented in MLPack
18:29
<
marko >
but it is still solid
18:31
<
zoq >
The field is so broad it's tricky to keep up.
18:31
<
jeffin143[m] >
And it also grows
18:31
<
jeffin143[m] >
So consuming it becomes
18:31
<
jeffin143[m] >
More difficult
18:32
<
jeffin143[m] >
Each day they is some new things Ppppin
18:32
<
jeffin143[m] >
Popping up
18:32
<
jeffin143[m] >
I have made a list to read those and it has only grew , can't keep even track of it
18:35
<
marko >
its important that there is progress
18:35
<
marko >
keep up the good work
18:35
<
marko >
we keep in touch
18:35
<
zoq >
see you around
18:36
marko has quit [Remote host closed the connection]
18:57
<
abernauer[m] >
IT job check scams are still a thing during pandemic times.
19:58
ImQ009 has quit [Quit: Leaving]