09:55 < rf_sust2018> hi.im rafee from bangladesh. currently studying physics and computer science in SUST. i want to work on Java bindings for mlpack.let me know if anyone is assigned to it.

11:25 < rf_sust2018> rcurtin: hello there

12:16 < zoq> rf_sust2018: Hello, no one is currently working on Java bindings, so feel free to pick that up; http://knife.lugatgt.org/pipermail/mlpack/2018-February/003577.html should be helpful, make sure to check out the responses as well.

12:18 < zoq> rajeshdm9: We will take a look at proposals once the application phase opens, the review process takes time, so please be patient.

16:13 < s1998_> zoq: w.r.t linear.hpp, InputDataType& InputParameter() { return inputParameter; }, what is the purpose of this funcion ? Or if possible, is there any resource(or keyword that needs to be googled) that you can point me to ?

16:48 < zoq> s1998_: InputParameter() is there to get/set/store the input parameter, e.g. in a model the output of layer n is stored in layer n + 1, InputDataType is data type e.g. arma::mat

18:00 < s1998_> zoq: what is the size limit of datasets that can be stored for mlpack/models ?

18:06 < zoq> hm, I think the github limit is something around 150 MB, but we should see if we can zip the archive, and extract it as a build step.

18:07 < s1998_> okay :)

18:33 < Atharva> Is there a way to run the style checks locally on my system?

18:35 < zoq> Atharva: Yes, get https://github.com/mlpack/jenkins-conf/blob/master/linter/lint.sh and do something like: ./lint.sh --root . --reports cpplint.xml

18:35 < zoq> You also need the other files: https://github.com/mlpack/jenkins-conf/tree/master/linter

18:36 < Atharva> Okay, I will try this, thanks.

18:37 < zoq> Let me know if you run into any issues.

18:37 < Atharva> Yeah sure.

19:08 < Atharva> zoq: The cpplint.xml file is supposed to have the results, right?

19:08 < zoq> correct

19:09 < Atharva> Okay, so ran the command and that xml file is blank

19:09 < zoq> I guess, the code does have style issues?

19:10 < Atharva> Yes it does still have style issues according to the online check

19:11 < zoq> hm, okay what if you remove line 65 and the '| \' from line 64?

19:16 < Atharva> Now, the xml file is still blank but I get some output in the terminal. It informed a lot of times about ignoring and skipping files which weren't cpp, hpp.

19:16 < Atharva> On the last line it says "Total erros found: 15"

19:16 < Atharva> errors*

19:17 < Atharva> So I guess it works but how do I know where the style errors are?

19:19 < zoq> If you remove the "skipping files" messages (e.g. using grep), do you see the issues?

19:19 < zoq> Looks like there is some problem with the xml conversion on your system, whcih I think is okay, since you don't need the xml format

19:22 < Atharva> Sorry, I don't understand what you mean by removing the "skipping files" messages, can you point me to some example? Do you mean to say that I can use grep to omit certain terminal outputs?

19:24 < zoq> yes, you can adpat "grep -v 'Consider using rand_r'" which ignores lines that complain about using rand_r.

19:32 < Atharva> Okay, so there is no information about the file with the errors in the log. It now directly shows the last line.

19:33 < Atharva> This is very useful, using grep to control terminal output. :)

19:34 < zoq> do you run the command inside the build folder or the root folder?

19:36 < Atharva> zoq: root folder

19:38 < zoq> Can you make sure that: https://github.com/mlpack/jenkins-conf/blob/6d46d82b1ca85e1f76e7c75f4e85eaba7b27a6a2/linter/lint.sh#L37-L45 contains the file with the style issues?

19:39 < zoq> Not sure what the issue here is, it looks like it's not picking up the file, so we have to go through the script step by step.

19:39 < apoorv> zo

19:40 < apoorv> Hi zoq this is apoorv from india i want to work on reinforcement learning if it were not assigned to someone else

19:44 < apoorv> zoq could you please help me that how can we implement double dqn algorithm

19:44 < zoq> apoorv: Hello, I guess you are talking about the project in the context of GSoC, the application phase isn't open or closed yet, so we haven't made any decision at this point.

19:45 < Atharva> zoq: Okay, I want to check knn_test in main_tests but that file isn't there.

19:46 < Atharva> I tried manually adding it but it didn't work.

19:47 < zoq> apoorv: A good first step is to get familiar with the existing codebase, there is already an implementation of Double DQN, so afterward we should talk about extensions if you like to work on that part.

19:47 < zoq> Atharva: Interesting, let's see.

19:50 < apoorv> Thanks zoq i will start implementing the given codebase

19:50 < apoorv> And after we will discuss about its expansion in atri games.

19:51 < apoorv> Atari*🎂

19:51 < Atharva> zoq: Yeah, I will try doing all this again to make sure that I am not doing something wrong. I will let you know.

19:52 < zoq> Atharva: Do you see other main tests? On my system it's picking up everything in main_tests.

19:54 < zoq> apoorv: Okay, but note, as I pointed out Double DQN, is already implemented, but I think there are some features that could be added.

19:55 < apoorv> What features can we add more in double DQN

20:00 < Atharva> zoq: Sorry, I was making a mistake there, the file was getting checked, it's just that I made some corrections before that and there are actually no style errors now.

20:01 < Atharva> zoq: But this was the 16th error and it showed the details. What about the other 15 errors?

20:05 < Atharva> zoq: I mean that it does work as expected, i made a deliberate error and it did show the details now, but why doesn't it show the details for the 15 other errors that it's detecting?

20:27 < zoq> Atharva: that are the issue we filter

20:27 < zoq> Atharva: you can remove the grep commands, if you like to see the rest

20:28 < zoq> Atharva: but this issues are fals warnings

20:28 < zoq> Atharva: check out the irc log in case you missed my messages :)

20:29 < zoq> apoorv: one idea is Multisource Transfer Double DQN

20:30 < Atharva> zoq: I checked. Okay, so they are false warnings. Got it. Thanks and sorry for that mistake of mine.

20:35 < zoq> No worries

23:07 < Mina> Hi! I am here for GSoC. Is this the right place to be?

23:10 < zoq> Mina: Hello there, yes this is the place to be.

23:15 < Mina> Great! Since proposal submission starts soon I thought it'd be best to get in touch with the mentors see if there are any requirements or tasks I should be aware of.

23:17 < Mina> I am particularity interested in the "Reinforcement learning" learning project.

23:20 < zoq> I see, there are no requirements like code contributions. A good first step is to get familiar with the codebase since you are interested in the RL project, you should check out the code and the corresponding tests.

23:20 < Mina> I've read the wiki on Github and found it requires a working knowledge of neural networks and reinforcement learning. But also I've found that "The "necessary knowledge" sections can often be replaced with "willing to learn" for the easier projects, and for some of the more difficult problems, a full understanding of the description statement and some coding knowledge is sufficient."

23:22 < zoq> Right, for the RL project, some knowledge about neural network is definitely helpful, since there isn't enough time to cover everything over the summer. But an in-depth knowledge about the method is somewhat that could be covered by a willingness to learn.

23:25 < Mina> I am actually experienced in C/C++ but I still have very basic knowledge in Machine Intelligence in general, tho I intend to pursue it as a career path. So, my question is: do I have a chance in getting accepted despite not possessing the requirements yet?

23:28 < zoq> At this point, it's difficult to answer that, since as you pointed out the application phase hasn't started yet, but this is one of our most popular ideas, and we can probably only accept one student working on it.

23:36 < Mina> Completely understandable. Thanks for your help! I'd really appreciate it if you'd suggest a less popular idea that I may have better chance at and maybe guide me on how to prove some competance.

23:36 < Mina> *competence

23:44 < zoq> Honestly, you should pick the ideas you are interested in or propose an own idea; if you like to apply for less popular ideas I think a good indicator is the mailing list archive.

23:44 < zoq> In case you haven't seen it already: mlpack.org/gsoc.html and www.mlpack.org/involved.html might be helpful.

23:51 < Mina> I've seen them but thought I'd get in touch first. Anyway I am really interested in the project I mentioned and would gladly try to help even outside GSoC scope, so if there's nothing I can do to boost my chances I might as well do so.

23:56 < zoq> Showing us, that you are familiar with the codebase e.g. by writing an example, extending/improving a method, etc. is helpful not only when writing the proposal; the proposal should make clear what you like to do over the summer, what method could be reused, what method needs to be extended, etc.