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/
< sakshamb189[m]>
Hey kartikdutt18 do we have a meet right now?
< kartikdutt18[m]>
Hey sakshamb189 , Yes we do.
< kartikdutt18[m]>
For updates, I have completed the PreProcessor PR for YOLO model. I also wrote a function for visualization (plot / save) for bounding boxes and also wrote the Forward pass for YOLOv3 network. The next task is completing the loss function so that we can merge the YOLO PR.
< sakshamb189[m]>
alright so, are you implementing YOLOv3 first?
< kartikdutt18[m]>
Not really. I can complete that first if we want. I did that, so that I can understand how we will concatenate output for the loss function so I wrote the forward pass.
< sakshamb189[m]>
alright sure no problem so the YOLO class PR implements yolov1 right?
< kartikdutt18[m]>
Yes.
< kartikdutt18[m]>
Other than style fixes and verifying it's output it is ready to go.
< sakshamb189[m]>
and the loss funciton is required for the YOLO class PR?
< sakshamb189[m]>
function*
< kartikdutt18[m]>
Yes.
< sakshamb189[m]>
How are you planning to test it out? Similar to weight transfer we did before?
< kartikdutt18[m]>
Yes, since the layers are same i.e. composed of Conv, Linear and BatchNorm we can do that very easily.
< sakshamb189[m]>
do you think you would be able to finish the testing this week?
< kartikdutt18[m]>
I think so, the difficult part was to figure out the working of loss function and I have moved on to implementing it. I think we should be able to merge the YOLO PR this week.
< sakshamb189[m]>
alright sounds good. Is there anything else we need to discuss?
< kartikdutt18[m]>
Could you take a look at the PreProcessor PR. I think that it is ready to go.
< sakshamb189[m]>
hmm I do not think we would be able to merge that before merging the actual YOLO model. Let me know if that is wrong
< rcurtin>
zoq: thanks for the email reminder, I had forgotten the video meetup was today :-D
< sakshamb189[m]>
I'll take a look by this week.
< kartikdutt18[m]>
Makes sense, We can merge them all together then, the Loss function, the preprocessor and the model.
< zoq>
rcurtin: Strange, I don't the the mail in my inbox.
< zoq>
rcurtin: But good it went out.
< rcurtin>
zoq: it's because fu-berlin.de still doesn't seem to like mlpack.org's new IP:
< sakshamb189[m]>
Alright, anything else apart from PreProcessor PR?
< rcurtin>
(host mail.fu-berlin.de[130.133.4.67] said: 451-161.8.32.26 is not yet authorized to deliver mail from 451-<mlpack-bounces@lists.mlpack.org> to <marcus.edel@fu-berlin.de>. Please try 451 later. (in reply to RCPT TO command))
< zoq>
rcurtin: ahh
< kartikdutt18[m]>
Nothing more from my side.
< sakshamb189[m]>
Let me know if you have any questions/blockers.
< sakshamb189[m]>
We can meet next week then. Hope you have a nice weekend ahead!
< kartikdutt18[m]>
Sure. Great, Have a great week.
< zoq>
rcurtin: Now I got the mail.
< rcurtin>
zoq: yeah, after the IP change, I spent a long time setting up email authentication mechanisms like DKIM and DMARC, etc.
< rcurtin>
I see that shrit[m]'s mailserver greylisted me, but I'm not totally sure why :) it looks like postfix resent both to fu-berlin.de and shrit.me and succeeded though
< rcurtin>
so long as mailing list messages are being successfully received by gmail and other major providers, there aren't any huge issues (so if someone has a gmail account on the mlpack list and *didn't* yet receive the meetup notification... let me know! :))
< rcurtin>
those of us who use our own mailservers are strongly in the minority :)
< rcurtin>
nice that fu-berlin.de (and, it looks like, iu.edu) run their own mailservers; some years ago Georgia Tech switched from running their own to simply paying Microsoft to run it
< zoq>
rcurtin: Setting a mailserver up right, is not the easiest.
< rcurtin>
yes, every few years I go to debug my configuration and find all kinds of things that were wrong that somehow I didn't catch last time
< rcurtin>
DMARC for instance is a giant nightmare to set up fully... I did not go through that whole process
< zoq>
I guess you "just" go the postfix way, so no Dovecot on top?
< rcurtin>
dovecot is set up for some other users on the system, but personally I just run mutt directly on the system, so no need for dovecot for me :)
< zoq>
That's what I thought :)
< KimSangYeon-DGU[>
kartikdutt18: sakshamb189 Sorry for being away from the meeting. Recently, I felt sick, slept early and didn't wake up even several alarms... I checked the progress you guys discussed. Sorry about that again.
< KimSangYeon-DGU[>
* kartikdutt18: sakshamb189 Sorry for being away from the meeting. Recently, I felt sick, slept early and couldn't wake up even several alarms... I checked the progress you guys discussed. Sorry about that again.
< kartikdutt18[m]>
Hey KimSangYeon-DGU , No worries. I hope you are feeling better. Take care.
ImQ009 has quit [Read error: Connection reset by peer]
ImQ009 has joined #mlpack
< zoq>
I won't be able to join the meeting today.
ImQ009 has quit [Read error: Connection reset by peer]
ImQ009 has joined #mlpack
ImQ009 has quit [Read error: Connection reset by peer]
ImQ009 has joined #mlpack
kristjansson has joined #mlpack
< rcurtin>
zoq: no problem, if you want, we could send a poll to change the time, if this time is continually bad for you and you'd like to join
bnbdim has joined #mlpack
bnbdim has quit [Remote host closed the connection]
ImQ009 has quit [Quit: Leaving]
< rcurtin>
shrit[m]: just wanted to check in, how is the cereal debugging going?
< zoq>
rcurtin: yes, let's do another poll and see what people think
< rcurtin>
sure, there were only 3 to 5 of us this time, so maybe it is not the best time
< zoq>
Will send out a mail with the poll.
< rcurtin>
awesome, thanks!
< shrit[m]>
rcurtin, not that good, I have tried to use valgrind, I was not able to find the cause of the error
< shrit[m]>
@rcurting if you have any idea that would be very helpful