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/
< zoq>
rcurtin: I thought this only effects Jenkins, but looks like mlpack.org as well.
< rcurtin>
zoq: yeah, I am really confused what happened here
< rcurtin>
no signs of foul play
< rcurtin>
my network is not very fast, so copying the backups back into place could take a while
< rcurtin>
I'll prioritize the websites, hopefully only a couple hours for those
< zoq>
I think it's not super critical.
< zoq>
I wondering are the files gone?
< rcurtin>
well the websites are pretty critical I think :)
< zoq>
Or is this some configuration issue?
< rcurtin>
the directories /var/www/www.mlpack.org/ and /var/www/www.ensmallen.org/ are empty
< zoq>
ohh
< rcurtin>
but I don't know why---none of the other websites had this issue, only the ones controlled by jenkins
< zoq>
Did we enable the jenkins auto update?
< rcurtin>
for jenkins itself, it shouldn't be enabled
< rcurtin>
but for the plugins maybe
< rcurtin>
ok, got ensmallen.org back online
< zoq>
nice
< AlexNguyenGitter>
@zoq Thabks for the answer the other day, I can find the command through 'mlpack_test --help' now
< AlexNguyenGitter>
And also I have a small question, when I write test case or new method, how should I debug it (i.e. Print the current generated matrix) ? Is there any flag to show debugging printing?
yac87 has joined #mlpack
< yac87>
Hello o/
< zoq>
yac87: Hello there.
< yac87>
I'm trying to figure out howto serialize mlpack::gmm::GMM - when I use cereal Archive ar(gmm,1) I get "cereal found more than one compatible output serialization function for the provided type and archive combination"
< zoq>
AlexNguyenGitter: To print Log::Debug you need to set mlpack::Log::Debug.ignoreInput = false; e.g. at the start of the test.
< yac87>
even when I specialize like this: CEREAL_SPECIALIZE_FOR_ALL_ARCHIVES(mlpack::gmm::GMM, cereal::specialization::member_serialize);
< yac87>
to clarify I just want to save/load GMM to file and I noticed it has serialize() method which looks like what I want but I don't know how to invoke it properly /o\
< zoq>
yac87: data::Save("model.xml", "gmm", gmm, false); should do the trick
< zoq>
yac87: data::Load("model.xml", "gmm", gmm); to load
< yac87>
Oh I though it only works for arma data types... thanks ᕕ( ᐛ )ᕗ
< yac87>
BTW mlpack.org is down for me... or rather it's up but serving 404 and weird half-loaded pages. I assume it's known, but just to be sure...
< zoq>
Yes thanks for the info, Ryan is currently working on getting it back online.
< yac87>
zoq oh it's so nice... modern C++ isn't even that bad with beautiful libs like this <3
< zoq>
True :)
< yac87>
hehe, True (python) vs true (c/c++) is costing me extra typing :-p Anyway, thanks a lot for the help and for the library itself, it's awesome!
pradkrish has joined #mlpack
< RishabhGarg108Gi>
Hey all, I wanted to know if there is any facility to freeze the weights of some layers in mlpack, if we don't want them to get changed during backpropagation. For example, like we freeze the weights of the top layers in transfer learning.
< zoq>
RishabhGarg108Gi: There is no such option, but would be interesting to have.
< abernauer[m]>
Yeah that would be a solid featur to implement agree.
< abernauer[m]>
* Yeah that would be a solid feature to implement agree.
< RishabhGarg108Gi>
Yes, It would be a big step. This can also open doors and we can provide ready to use implementations of various state of the art CNN models like Inception, Residual networks etc.
< rcurtin>
yac87: zoq: yeah, restoring from backups right now... the backup should be transferred over within 30-45 minutes
< RishabhGarg108Gi>
I think I will open an issue for it.
AakashkaushikGit has quit [Ping timeout: 260 seconds]
RakeshBhugraGitt has quit [Ping timeout: 260 seconds]
geek-2002Gitter[ has quit [Ping timeout: 260 seconds]
RakeshBhugraGitt has joined #mlpack
geek-2002Gitter[ has joined #mlpack
VSaicharanGitter has quit [Ping timeout: 260 seconds]
ChaithanyaNaik[m has quit [Ping timeout: 260 seconds]
_slack_mlpack_U0 has quit [Ping timeout: 260 seconds]
AnushKiniGitter[ has quit [Ping timeout: 260 seconds]
turska79Gitter[m has quit [Ping timeout: 260 seconds]
VSaicharanGitter has joined #mlpack
turska79Gitter[m has joined #mlpack
AnushKiniGitter[ has joined #mlpack
ChaithanyaNaik[m has joined #mlpack
AakashkaushikGit has joined #mlpack
_slack_mlpack_U0 has joined #mlpack
< AakashkaushikGit>
Hey I am stuck at the copy_visitor and i can't actually understand what `CopyVisitor<CustomLayers...> ` is i looked at `layer_types.hpp` but i can't understand what that means in the context of `LayerTypes`
< AakashkaushikGit>
Basically I want to implement the same without using boost visitors.
< AakashkaushikGit>
in the ffn class
< rcurtin>
ok, got the website back up...
< yac87>
\o/
pradkrish has quit [Ping timeout: 245 seconds]
< zoq>
rcurtin: Nice!
ImQ009 has quit [Quit: Leaving]
yac87 has quit [Remote host closed the connection]