rcurtin_irc changed the topic of #mlpack to: mlpack: a scalable machine learning library (https://www.mlpack.org/) -- channel logs: https://libera.irclog.whitequark.org/mlpack -- NOTE: messages sent here might not be seen by bridged users on matrix, gitter, or slack
TrinhNgo[m] has quit [*.net *.split]
LolitaNazarov[m] has quit [*.net *.split]
sdev_7211[m] has quit [*.net *.split]
_slack_mlpack_34 has quit [*.net *.split]
_slack_mlpack_13 has quit [*.net *.split]
TathagataRaha[m] has quit [*.net *.split]
jeffin143[m] has quit [*.net *.split]
Gulshan[m] has quit [*.net *.split]
M7Ain7Soph77Ain7 has quit [*.net *.split]
rcurtin[m] has quit [*.net *.split]
AbhishekMishra[m has quit [*.net *.split]
halfy has quit [*.net *.split]
_slack_mlpack_13 has joined #mlpack
halfy has joined #mlpack
_slack_mlpack_34 has joined #mlpack
M7Ain7Soph77Ain7 has joined #mlpack
jeffin143[m] has joined #mlpack
Gulshan[m] has joined #mlpack
TrinhNgo[m] has joined #mlpack
AbhishekMishra[m has joined #mlpack
rcurtin[m] has joined #mlpack
TathagataRaha[m] has joined #mlpack
<SuvarshaChennare> Yes On all
<SuvarshaChennare> Yup I followed everything
LolitaNazarov[m] has joined #mlpack
sdev_7211[m] has joined #mlpack
_slack_mlpack_14 has quit [*.net *.split]
SaiVamsi[m] has quit [*.net *.split]
HrithikNambiar[m has quit [*.net *.split]
KumarArnav[m] has quit [*.net *.split]
AbhishekNimje[m] has quit [*.net *.split]
AbhijayMitra[m] has quit [*.net *.split]
zoq[m] has quit [*.net *.split]
DirkEddelbuettel has quit [*.net *.split]
shrit[m] has quit [*.net *.split]
gitter-badgerThe has quit [*.net *.split]
_slack_mlpack_14 has joined #mlpack
gitter-badgerThe has joined #mlpack
shrit[m] has joined #mlpack
zoq[m] has joined #mlpack
HrithikNambiar[m has joined #mlpack
SaiVamsi[m] has joined #mlpack
AbhijayMitra[m] has joined #mlpack
KumarArnav[m] has joined #mlpack
DirkEddelbuettel has joined #mlpack
AbhishekNimje[m] has joined #mlpack
<zoq[m]> Can you remove the build folder an try again without manually setting all the path parameters, but with `-DDOWNLOAD_DEPENDENCIES=ON`?
<zoq[m]> We have to update the tutorial, and mention the new download option.
<SuvarshaChennare> > Can you remove the build folder an try again without manually setting all the path parameters, but with `-DDOWNLOAD_DEPENDENCIES=ON`?
<SuvarshaChennare> I still have to keep the armadillo and openBLAS path parameters right?
<zoq[m]> > <@suvarshachennareddy-614749f56da0373984860afa:gitter.im> > Can you remove the build folder an try again without manually setting all the path parameters, but with `-DDOWNLOAD_DEPENDENCIES=ON`?
<zoq[m]> > I still have to keep the armadillo and openBLAS path parameters right?
<zoq[m]> No, that should be handled by the downloader, shrit implemented it, so correct me if I'm wrong.
<zoq[m]> >
<shrit[m]> No, I do not think you need to specify any dependency
<jonpsy[m]> zoq: say4n jjb so we're meeting at the zoom link right?
<say4n[m]> works for me
<zoq[m]> Actually I was wondering if we should move it to tomorrow, unless today works better for everyone?
<jonpsy[m]> oh.. others have joined. np let's meet next week
<jonpsy[m]> unless you're free now
<zoq[m]> <jonpsy[m]> "oh.. others have joined. np let..." <- Okay, if others joined, good, I will read the docs.
<heisenbuugGopiMT> Hey @shrit:matrix.org I shifted from fedora to Ubuntu and I am facing a lot of issues so I am not able to use my machine. I hope to solve everything by tonight and will complete the work on test by tomorrow end of the day.
<shrit[m]> Perfect let know if you face any issue
<heisenbuugGopiMT> Yup, also do you have any idea why my fans might be running on max speed since I installed Ubuntu? It was fine in Fedora.
<heisenbuugGopiMT> Lets hope I can solve this issue...
<shrit[m]> Use htop to see what is eating memory
<heisenbuugGopiMT> Also since I installed CUDA, I am not able to put my system on sleep...
<heisenbuugGopiMT> So I am guessing it;s due to my GPU drivers only
<shrit[m]> try reboot, each time I installed a new nvidia driver it requires a reboot
<heisenbuugGopiMT> I did many times...
<shrit[m]> Also use gnome or some light desktop environment
<jonpsy[m]> <zoq[m]> "Okay, if others joined, good..." <- So the meeting summary is in the invite itself. James suggested that it'd be better to send out an agenda + brief + links of the meetings beforehand so I've done that. I expect we'll make good progress on the PR review by next week :). Send the link , you're okay with Monday weekly right?
<jonpsy[m]> s/Send/Sent/
<shrit[m]> what environment you have?
<shrit[m]> this is really high
<shrit[m]> do you have nvidia gpu?
<heisenbuugGopiMT> Yup
<heisenbuugGopiMT> 1050Ti
<shrit[m]> try nvtop to see your consumption
<shrit[m]> on nvidia
<heisenbuugGopiMT> ```
<shrit[m]> Oops, why I do not use Ubuntu
<shrit[m]> I suppose you have a newer version of nvidia on you mahcine
<heisenbuugGopiMT> What are you using then?
<shrit[m]> I use arch,
<shrit[m]> do you know how to use tailing windows managers
<shrit[m]> ?
<heisenbuugGopiMT> Nope
<shrit[m]> What is the prime profile for nvidia setting?
<heisenbuugGopiMT> Are you talking about this one?
<heisenbuugGopiMT> Performace Mode
<heisenbuugGopiMT> It's in performance mode
<heisenbuugGopiMT> This is result of htop
<shrit[m]> Your gnome shell is eating a lot of CPU in addition to your Xorg server
<shrit[m]> there is something wrong,
<shrit[m]> but I suppose that is the fan of the GPU that is running
<shrit[m]> if you can change from the performance mode that would be a good start
<heisenbuugGopiMT> Should I change it to intel?
<shrit[m]> You have an intel GPU?
<shrit[m]> that will make the CPU fan run similarly
<heisenbuugGopiMT> There are three options here
<heisenbuugGopiMT> NVIDIA On-Demand
<heisenbuugGopiMT> Intel(Power Saving Mode)
<heisenbuugGopiMT> NVIDIA (Performance mode)
<shrit[m]> Put it on demand
<heisenbuugGopiMT> Done, I need to restart to apply setting, I will be back in a minute.
<heisenbuugGopiMT> Nope, its the same
<heisenbuugGopiMT> Do you think it's better to change the OS?
<heisenbuugGopiMT> I don't have any strong preference...
<shrit[m]> the issue is not in the OS
<shrit[m]> it is in the Xorg server and the desktop environemt
<heisenbuugGopiMT> Shall I try changing desktop environment then?
<shrit[m]> Yeah, try have something that is light
<heisenbuugGopiMT> Any suggestions?
<shrit[m]> for laptop, I would use somehting that is a tailing window manager, like DWM and I3,
<heisenbuugGopiMT> Also since we changed to on demand I can see a lag on my desktop. Keyboard strokes and mouse movements are lagging
<shrit[m]> if we find it hard to use these, use something like mate or gnome 2 or anything within this range
<shrit[m]> okay that is strange, do you know what nvidia driver you used on Fedora?
<heisenbuugGopiMT> Nope, I just installed fedora and never needed to do anything...
<heisenbuugGopiMT> I don't see a lag now. I had fractional scaling enabled maybe thats why it was lagging?
<shrit[m]> I do not know
<heisenbuugGopiMT> The lag is gone atleast but the fans are still rocking...
<heisenbuugGopiMT> I am installing i3.
<heisenbuugGopiMT> I will shift to it as soon as it's done and lets see what happens there
<shrit[m]> I3 need to be configured
<shrit[m]> with no conf you can get an empty screen at start
<heisenbuugGopiMT> I can follow this if you think its good
<shrit[m]> zoq: do you remember which example in our example repo used the `stb`to load images?
<shrit[m]> heisenbuug (Gopi M Tatiraju): that is good I think
<heisenbuugGopiMT> Okay, its working and its silent...
<heisenbuugGopiMT> But I need to learn using I3
<shrit[m]> not that hard, once learned it will be easier to deal with than other
<heisenbuugGopiMT> Yes, wow...
<heisenbuugGopiMT> I am already loving it
<heisenbuugGopiMT> thank you soo much
<shrit[m]> zoq: if I have an image with 3 channel how to use CNN in mlpack, that was my question more precisely
<shrit[m]> because the CNN constructor does not specify the channel
<shrit[m]> otherwise can we flatten the 3 channels into one matrix? but not sure how the filter are going to be applied
<zoq[m]> You are right that you have to flatten the 3 channels, [R, R, ... R, G, G, ..., G, B, B, ..., B].
<zoq[m]> The cnn/conv layer will use the filter and input size to 'reconstruct' the input shape.
<zoq[m]> So you can use the architecture the same way as you would use it with a single channel.
<shrit[m]> Okay, So technically, the input size is the channel and the CNN constructor is going to take care of everything
<shrit[m]> if channel is 1 then input is 1 and etc..
<shrit[m]> so I do have to do any manual handling, the issue with the example it is using csv
<zoq[m]> yep, first first parameter number of input channels.
<shrit[m]> so I suppose that the matrix are already flattened?
<shrit[m]> if I parsed an image using the `stb` do I have to do manual flatten ?
<zoq[m]> Yes, in this case that is correct, you could build with stb support that will do the same thing for you.
<shrit[m]> okay so stb flat everything,
<zoq[m]> let me verify that quickly
<zoq[m]> yes
<shrit[m]> ah cool, image are already 8 bit integer
<shrit[m]> perfect 👍️
<zoq[m]> Yes, this is what stb returns.
<zoq[m]> Depending on the input type you pass to the load function it will get converted at the end.