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/
ImQ009 has joined #mlpack
travis-ci has joined #mlpack
< travis-ci> UtR491/ensmallen#1 (moead_exp - 9bb28a4 : UtR491): The build failed.
travis-ci has left #mlpack []
< R-Aravind[m]> @anjishnu:matrix.org: I think I'm hit a wall on the backward method of MultiLabelMarginLoss. The results of the one from pytorch and the math doesnt makes any sense. So far the forward pass works well and gives the correct outputs.
< anjishnu[m]> Can you create a notebook similar to the ones I have for the other loss functions and share it here? Then I can take a look at it.
< jeffin143[m]> Hey everyone , I could help Mlpack apply for mlh fellowship, do anyone have any objection about me being single point of contact with mlh organizers ?
< jeffin143[m]> Also would appreciate to come up with any ideas if you have in your to do list
< jeffin143[m]> Any new features or for that instance any new algorithms you want to see in mlpack
< jeffin143[m]> You can help in reviewing the pr while they implement it :)
< zoq> jeffin143[m]: Just glanced over the program at this point, looks like students have to pay some fee to get there proposal reviewed?
< zoq> Also, do we have to come up with tasks/projects?
< zoq> Looking at some of the tasks from the mail, I'm not sure it's enough work to spend 3 months on it?
< jeffin143[m]> zoq (@freenode_zoq:matrix.org): yes they have to pay 7 Dollars or max of 25 dollars
< jeffin143[m]> But it's not mandatory , it's kind of contribution
< jeffin143[m]> Also it is not mandatory to work on a single project
< jeffin143[m]> They just have to keep on contributing
< jeffin143[m]> It could be distributed across repo and also distributed across features
< jeffin143[m]> We have to come up some kind of roadmap*
< jeffin143[m]> Like bugs to be fixed , features to be implemented , things to be improved , refactoring
< jeffin143[m]> We could have it as a project board and then let students decide what to pick
< zoq> Looks like we don't have to provide a link to that now?
< jeffin143[m]> zoq (@freenode_zoq:matrix.org): no we don't have to , it should be there in our backlog , so that after students get selected they know what to do
< jeffin143[m]> Here students don't submit like full fledged proposal
< jeffin143[m]> They submit similar work experience like c++ + ml project's and mlh decides then
< zoq> jeffin143[m]: I see, happy to help out.
< jeffin143[m]> Yes probably coming Thursday would be the first one of the week let's
< jeffin143[m]> Come up with some ideas so that we can have it in our to do list if by chance we get accepted
< jeffin143[m]> Even if we didn't get accepted may be any new contributors can get involved
< zoq> Sounds good, partially you can already search for "good first issue" and "help wanted" in the issues list.
< jeffin143[m]> Yes :)
ImQ009 has quit [Quit: Leaving]
< rcurtin> jeffin143[m]: I read a bit more about the idea, it's a bit concerning that MLH assigns the mentor, and so essentially the folks selecting who is qualified to work on mlpack don't really know anything about the mlpack project itself (if I understand right)
< rcurtin> I think this can still work alright if we have some contributors helping the MLH mentors and students understand what our coding guidelines are and doing rigorous code reviews, etc.
< rcurtin> (for what it's worth... mlpack is far more detailed with reviews than other projects I have seen, and personally I think that is a good thing that helps keep quality high :))