[mlpack] Help Needed : Q learning using Mlpack
Marcus Edel
marcus.edel at fu-berlin.de
Thu Feb 22 16:35:32 EST 2018
Hello Rohan,
thanks for the update.
> One more request to you. Please brief me any procedure by which we can add these
> tags to our code. Presently I have aliased the command in the bashrc of my
> system. It will be more convenient for the users to run the code directly with
> g++.
One option that could work depending on your system is:
g++ example.cpp -std=c++11 `pkg-config --libs mlpack`
In the end, there is no general solution and alias might work in your case, but
it would look different if you like to use mlpack in conjunction with another
library.
> Presently I am documenting my first RL code on mlpack on my blog page. After
> this I will try simple policy gradient method on cartpole . I would like to know
> Shangton's and Ryan's views on it. Please let me know if you wish to give any
> suggestions.
Sounds good, we will look at the PR once we get a chance.
I hope this is helpful, let me know if I should clarify anything.
Thanks,
Marcus
> On 21. Feb 2018, at 05:30, Rohan Raj <rajrohan1108 at gmail.com> wrote:
>
> Hello Marcus,
>
> Thanks for being the saviour. The code ran smoothly. Please let me know how to render the output window of the gym.
>
> Previously I was running the codes without #include <boost/test/unit_test.hpp> . However same error was encountered . What I see here is that you had added an executable in the command. I sincerely request you to add the same executable in your comment on the issue <https://github.com/mlpack/mlpack/issues/562#issuecomment-204810912>.
>
> One more request to you. Please brief me any procedure by which we can add these tags to our code. Presently I have aliased the command in the bashrc of my system. It will be more convenient for the users to run the code directly with g++.
>
> Presently I am documenting my first RL code on mlpack on my blog page <https://rohanrajblogs.blogspot.in/>. After this I will try simple policy gradient method on cartpole . I would like to know Shangton's and Ryan's views on it. Please let me know if you wish to give any suggestions.
>
> Thanks
>
> Rohan Raj
> Indian Institute of Technology Guwahati
> Assam , India
> Phone : +91 8723990557, +91 8651776581
>
>
>
>
> On 21 February 2018 at 01:57, Marcus Edel <marcus.edel at fu-berlin.de <mailto:marcus.edel at fu-berlin.de>> wrote:
> Hello Rohan,
>
> Either you can remove #include <boost/test/unit_test.hpp> or link against the
> boost unit test framework e.g. clang++ basic.cpp /path/to/lib/libmlpack.so
> -I/path/to/mlpack/include/ -larmadillo -lboost_unit_test_framework -std=c++11
>
> Let me know if that solves the problem.
>
> About Log::Debug see http://www.mlpack.org/docs/mlpack- <http://www.mlpack.org/docs/mlpack->
> git/doxygen/classmlpack_1_1Log.html#afd147ffc8dcfdceaccad2fa1dde00e30 for more
> information, it's bascially like std::cout but it will only print the output if
> you build mlpack with DEBUG=ON.
>
> I hope this is helpful, let me know if I should clarify anything. And a nice
> blog post would be awesome.
>
> Thanks,
> Marcus
>
>> On 20. Feb 2018, at 18:50, Rohan Raj <rajrohan1108 at gmail.com <mailto:rajrohan1108 at gmail.com>> wrote:
>>
>> Hello Shangtong and Marcus,
>>
>> I am Rohan Raj,aka luffy 1996 . I have earlier expressed my interest in reinforcement learning module for GSOC 18.
>>
>> Presently I am trying to implement policy gradient method using mlpack. Before diving into this problem I planned to run a simple q learning problem using Shangtong's last years code. However I was facing errors similar to this issue <https://github.com/mlpack/mlpack/issues/562>. I
>> tried implementing zoq's command by referencing the library and armadillo to g++, but still do not have any success.
>>
>> This might be an elementary mistake from my side in the code. However , helping me out by rectifying it will be a great help as I am new to mlpack. I am stuck here for a day. I would like to mention that I will document a blog , once this works (incase it hasn't been written.) I also have doubts regarding Log::Debug and its use in the code written by Shangton in the qlearning <https://github.com/mlpack/mlpack/blob/master/src/mlpack/tests/q_learning_test.cpp> test file. It will be great if any one explains me the use of Debug.
>>
>> The codes are available on my github <https://github.com/luffy1996/policy-gradient> page. basic.cpp is the code which I am mentioning in this email. I will add the README once the code works and I will document the blog .
>>
>> Points to note are:
>> 1) The reinforcement learning test passed success on my system.
>> 2) I have added the lib and include as said by zoq in the comment of issue <https://github.com/mlpack/mlpack/issues/562> still the similar error exists.
>>
>> I sincerely request you not to broadcast this message on mlpack lists. I am planning to document it once this code works. :)
>>
>> I will be grateful to your help.
>>
>> Thanks,
>>
>> Rohan Raj
>> Indian Institute of Technology Guwahati
>> Assam , India
>> Phone : +91 8723990557, +91 8651776581
>>
>>
>>
>> ᐧ
>
>
> ᐧ
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://knife.lugatgt.org/pipermail/mlpack/attachments/20180222/057e07d2/attachment.html>
More information about the mlpack
mailing list