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/
< rcurtin>
huh, different version of doxygen on polar?
< rcurtin>
1.8.13 is in use on mlpack.org; I remember working hard to get that specific version, since our scripts to deploy the website depended on that
< rcurtin>
I've also dropped the number of executors on master to 2 (don't want to cause too much slowdown on that system; it's used for a lot)
< rcurtin>
I'll add another system or two in the next couple of days (didn't get to it today)
< zoq>
version812: But let us know if there is anything that we should clarify.
version812 has quit [Ping timeout: 245 seconds]
jeffin143 has joined #mlpack
< jeffin143>
walragatver[m] birm[m] : are you both there ?
< walragatver[m]>
jeffin143: Yeah I am here
< walragatver[m]>
So jeffin. Before we start I want to know where are you currently staying?
< jeffin143>
I live in Bhopal , Madhya Pradesh , India
< walragatver[m]>
Okay great.
< jeffin143>
walragatver[m] : the only agenda from my side is : About the repo ( To track development status ) : I am not sure how should we go ahead, ask rcurtin to make a repo in mlpack and then code there or I make it as a personal repo and then transfer under mlpack later
< jeffin143>
just like shrit did it last year, If i am not wrong !
< jeffin143>
Then only agenda for today from my side is*
< walragatver[m]>
Last year sreenik did something on similar basis. If I am wrong. Anyways, I won't mind the either way unless we are on track. You can go whichever suits you the best.
< walragatver[m]>
Just let me know the final decision you make.
< jeffin143>
I am also ok either ways will have a talk with rcurtin over this and will inform you :)
< walragatver[m]>
Okay.
< walragatver[m]>
So, I went through your timeline looks good.
< jeffin143>
I updated the docs with timeline and the snippet of multi threaded code , I hope you would have glanced it, The multi threaded logs string instead of summary but we can adapt that
< jeffin143>
Should we support it out of the box ?
< jeffin143>
I mean sklearn allows you to plot cm without using matlplot lib from inside the bundle so should we support in mlpack too ? This is a header file wrapper only
< walragatver[m]>
But your first target is to write the CMake code for the repo. Do you think you can complete it till 7 june?
< walragatver[m]>
<jeffin143 "I mean sklearn allows you to plo"> Okay just a sec
< jeffin143>
walragatver[m] : I dont have prior experience with cmake , I was involved in understanding proto and tensorflow's backened to a great extend and I am comfortable now, Currently I am studing the linking part of protobuf and other cmake things So I am assuming I have 8-9 days before coding start so I can learn enough to setup repo and compile and
< jeffin143>
link properly
< jeffin143>
Also I thanks birm[m] is really good with cmake , We can seek his help in getting the repo done, I have seen him reviewing many pr which involved cmake changes and hence I guess we can get that done
< walragatver[m]>
Okay fine.
< walragatver[m]>
Okay great then.
< walragatver[m]>
Also there is one thing I want to discuss.
< walragatver[m]>
Yeah that conversion was not present in the code in you shared with the timeline so I thought you might have missed it.
< walragatver[m]>
Glad that you have went through it.
< jeffin143>
Noo , I just shared the mutex part which were referencing that day and were looking for some article on that
< walragatver[m]>
Yaa.
< jeffin143>
I was comfortable with the proto part, I was not so with mutlthreading part and hence I wrote a small mockup , but debugging on the way I just made it more concrete. I will add other snippets on the way
< walragatver[m]>
This event writer class logs event protos. I think we can also follow the same mechanism.
< walragatver[m]>
I am just discussing this so that we remain on same page.
< walragatver[m]>
I don't know whether you have gone through the same thing.
< walragatver[m]>
And yeah one more thing always refer tensorflow v1's code it's quite simple.
< walragatver[m]>
tf2 code is slightly different than tf1.
< jeffin143>
yes I have go through the same thing : ClosableQueue is equivalent to shareableQueue in the docs shared , EventFileWrite is LogWriter , The only difference is I gues in python thread need run function and it executes it, I have combined _Eventwritethread in Logwriter and the thread is writefunction()
< jeffin143>
I was referring to tf2 on most occasion and Since they introduced new way of logging in tf2
< jeffin143>
plugins based
< walragatver[m]>
<jeffin143 "plugins based"> Okay I will need to go through it.
< jeffin143>
yes previously they had summary.image with v1 but now everything calls tf.summary.write with tensor and add plugin name as `text` ,`image` ,`audio` and so on
< jeffin143>
But yes tf1 is much easier and for advanced logging such as multiple images or multiple tensor we can adapt tf2 thing but we can stick tf1 will also , Since they have mentioned they will support all the formats
< walragatver[m]>
<jeffin143 "walragatver : Also whats your ta"> Regarding this It's good thing to do, But let's keep it in the additional todo list. If we finish our project early we can work on it.
< jeffin143>
yeah lets not over estimate :)
< walragatver[m]>
<jeffin143 " But yes tf1 is much easier and "> Hmm thanks for the clarification I was just referring to the tf1 till now. I will go through tf2's code soon.
< RyanBirminghamGi>
I'm bad at keeping up with irc, but I'm here!
< RyanBirminghamGi>
let me read the logs!
< RyanBirminghamGi>
(at least it looks like gitter is up to date so I don't have to type on my phone :) )
< walragatver[m]>
birm: Glad that you are here. And no worries about the meet I will keep you updated.
< walragatver[m]>
<jeffin143 "Also I thanks birm is really goo"> birm: let us know your thoughts on this.
< walragatver[m]>
jeffin143: So I hope you know what would be your next target.
< RyanBirminghamGi>
Happy to help! You're looking to start a new repo primarily for log conversion, right? Or would it hold all things mlboard?
< jeffin143>
RyanBirminghamGi : For everything -> MlBoard as in logWriter major since as of now we are sharing frontend using tensorbaord
< jeffin143>
walragatver[m] yes , I have to learn cmake , Isn't it ?
< walragatver[m]>
<jeffin143 "walragatver yes , I have to lear"> Yeah.
< walragatver[m]>
Also feel free to share some blogs. I am also quite new to cmake :)
jeffin143 has quit [Remote host closed the connection]
< walragatver[m]>
birm: Anything you would like to discuss. I am done from my side today.
< jeffin143[m]>
Yes :) , it will help in reviewing
jeffin143 has joined #mlpack
< birm[m]>
Can you remind me when you had scheduled these meetings, if they are scheduled? I'm bad at keeping up with irc, so I'll set reminders for myself
< jeffin143[m]>
I will do it
< jeffin143[m]>
I will send a mail 2 hours in advance
< jeffin143[m]>
If that duration is ok :)
< birm[m]>
oh, ok! should be!
< birm[m]>
thank you!
< jeffin143>
No issues :)
< jeffin143>
Just a headsup , It is every Sunday and Wednesday 10:30 pm IST
< jeffin143>
But anyways I will send a remainder too
< walragatver[m]>
birm: As a double check. I will also remind you on the meeting day
< birm[m]>
I'll add a reminder, so I don't have to rely on your kindness alone ;)
< jeffin143>
Hahah Great
< walragatver[m]>
birm: jeffin143: Okay so that's it for today. Thank you for your time. Have a nice day and Take care.
< jeffin143>
Great, Bye GN
< birm[m]>
That should work until daylight savings messes up the relative timing :p
< RyanBirminghamGi>
Rest well!
< jeffin143>
Messaging Using two IDs birm[m] RyanBirminghamGi
< RyanBirminghamGi>
haha, one on gitter, one on my phone
< RyanBirminghamGi>
sorry for confusion
< RyanBirminghamGi>
let me see if I can do something about that
< jeffin143[m]>
Here I am with other id
petris has quit [Quit: Bye bye.]
petris has joined #mlpack
favre49 has quit [Remote host closed the connection]
jeffin143 has quit [Remote host closed the connection]