04:27 < travis-ci> mlpack/ensmallen#267 (conradsnicta-patch-1 - 0564e46 : Conrad Sanderson): The build passed.

04:27 < travis-ci> Change view : https://github.com/mlpack/ensmallen/commit/0564e46b9737

04:27 < travis-ci> Build details : https://travis-ci.org/mlpack/ensmallen/builds/528641607

08:05 < jenkins-mlpack2> Project docker mlpack nightly build build #316: STILL UNSTABLE in 3 hr 51 min: http://ci.mlpack.org/job/docker%20mlpack%20nightly%20build/316/

13:10 < heisenbug_> hey zoq, should I create pages like these for all other topics for MLPACK ZOO: https://heisenbuug.github.io/Long-Short-Term-Memory/

13:10 < heisenbug_> Like a short reading about the concept...

13:10 < heisenbug_> I also commented the same on that pr...

14:49 < rickllyxu> hello, Ryan, and everyone, I wonder why we allow to pass training data parameters to model constructor? and Why we allow user pass hyper-parameters in Train() function?

14:51 < rickllyxu> when I refactor code, I am gradually feeling that don't make sense

14:53 < rickllyxu> especially the latter question, I think Train() method should not take hyper-parameters. If it has to change hyper-parameter, a new model object should be construct.

14:57 < rickllyxu> If Train() method takes hyper-parameters, how could we know which hyper-para should to be used when training if the model object has saved hyper-paras as private members.

14:57 < sreenik> Nice read heisenbug (I never realised until today that you use 'heisenbug' and not 'heisenberg')

14:58 < rickllyxu> Of course, now the model do not save hyper-parameters. But I want to save it..

14:58 < rickllyxu> Ok

14:58 < rickllyxu> I'd like to read it

14:59 < sreenik> rickllyxu: https://heisenbuug.github.io/Long-Short-Term-Memory/ there you go

15:01 < rickllyxu> will, it is , ennn, not related to my problem. But thanks

15:02 < sreenik> But heisenbug, I don't know if you have finished with the article yet but the addition of a flowchart could be quite expressive; and a couple typos might need fixing

15:02 < sreenik> rickllyxu: Do you mean to retrain your model with a set of different hyperparameters

15:04 < rickllyxu> I mean, If we will retrain the model with different hyper-para, I think it is better to New a model object with desired hyper-para, rather than pass it through Train() method

15:07 < sreenik> rickllyxu: Not sure if I get what your exact issue is but you could always create a new model or use a copy constructor

15:08 < rickllyxu> Because now I want to save the hyper-para as private member variable of model object. However, If Train() also takes hyper-para, I don't know which hyper-para should be used.

15:11 < rickllyxu> Yeah, you're right. But I wonder the API thought.

15:12 < sreenik> Train() takes an optimizer object. Hyperparameters are the ones you define while creating the network model

15:14 < rickllyxu> yeah, I get it. But I think it is not reasonable that Train() takes hyper-paras.

15:15 < rickllyxu> However, some model object in MLPack allow to pass hyper-paras in Train()

15:19 < sreenik> rcurtin: I was occupied with my exams for a couple of weeks and was also looking into the theoretical portions of deep learning. Now I can finally resume work and finish the CLI implementation I was working on and fix the remaining issues regarding the parser. Where do you suggest the parser and its CLI should exist (like repo and subdirectory)?

15:20 < sreenik> Also to keep in mind is the fact that the parser should come handy in the tensorflow translator implementation

16:04 < heisenbug_> yea, I need to make some changes and idea of a Flowchart is good...I will also work on it sreenik

16:05 < heisenbug_> yea, its heisenBUG

18:01 < heisenbug_> do we have students for all the projects in gsoc or some projects are not included this year?

18:08 < jeffin143> Congratulations to all the contributors who got selected for the projects

18:08 < jeffin143> Good luck everyone :)

18:09 < heisenbug_> is the list out?

18:19 < heisenbug_> where can we check the results?

18:26 < haritha1313> Congrats to all those who got selected :)

18:27 < haritha1313> heisenbug_: Its on the GSoC website

18:29 < heisenbug_> can you send me the link? or do we have to visit individual organization page?

18:30 < Toshal> Thanks for accepting my proposal. I look forward to contribute. Also congratualations! to all others who got selected.

18:34 < Toshal> heisenbug_: This might be helpful. https://summerofcode.withgoogle.com/organizations/5868789747417088/#!

18:36 < heisenbug_> yea, I got that.

18:36 < heisenbug_> My proposal didn't got selected.

18:36 < heisenbug_> Congrats to you...

18:37 < heisenbug_> Congratulations to all those who got selected.

18:46 < RajatGarg97> Hi guys I am new joinee

18:47 < RajatGarg97> I want to contribute to mlpack. Are there any ongoing projects which I can contribute to?

18:57 < zoq> RajatGarg97: Hello, there are a couple of open issues.

19:01 < RajatGarg97> zoq: Hello, What procedure is followed here for merging pull requests and raising issues?

19:03 < favre49> Congrats to everyone who was accepted! Looking forward to the experience.

19:05 < zoq> RajatGarg97: https://github.com/mlpack/mlpack/blob/master/CONTRIBUTING.md should be helpful

19:14 < sooham> Hi mlpack devs, I am an undergraduate student at University of Toronto. I was about thinking of applying to GSoc this year but it would be incompatible with my schedule, however I still want to contribute towards one of the GSoc project ideas. Given today is the day for finalizing student applicants can I ask which projects are not taken

19:15 < zoq> sooham: Hello there, see https://summerofcode.withgoogle.com/organizations/5868789747417088/#5110501250433024

19:16 < favre49> hmmm is the website down? i can't seem to connect

19:17 < zoq> works for me

19:17 < favre49> hmmm probably a college wifi thing, thanks

19:41 < Toshal> zoq shkharJ saksham: Hi, I want to know how to move ahead. I saw that saksham is also going to work on GANS. Should I continue with Dual Optimizer idea?

19:49 < zoq> Toshal: Right, let us talk about each project and the timeline in the next days.