verne.freenode.net changed the topic of #mlpack to: http://www.mlpack.org/ -- We don't respond instantly... but we will respond. Give it a few minutes. Or hours. -- Channel logs: http://www.mlpack.org/irc/
vstsenroh has quit [Remote host closed the connection]
govg has quit [Quit: leaving]
govg has joined #mlpack
vivekp has quit [Ping timeout: 240 seconds]
vivekp has joined #mlpack
ImQ009 has joined #mlpack
ImQ009 has quit [Quit: Leaving]
ImQ009 has joined #mlpack
alsc has joined #mlpack
alsc has quit [Client Quit]
alsc has joined #mlpack
alsc has quit [Quit: alsc]
< ImQ009>
Hm
< ImQ009>
I've got a little problem whilst training my RNN. I get a crash in negative_log_likelihood_impl.hpp @ 39, because of currentTarget underflow
< ImQ009>
target(i) is 0
< ImQ009>
It seems to not like my labels being 0
< ImQ009>
When I set them all to 1, it's fine
< ImQ009>
It seems there's something I don't understand about the way I'm supposed to label my data..
< ImQ009>
When I look in the example RNN code there's this label post-process thing, which adds 1 to all labels... Hmmm
< ImQ009>
I don't think that's something I should be doing though. I tried it and I get the same kind of crash - out of bounds
< ImQ009>
Okay, I think I got it, got what I got wrong
< ImQ009>
I forgot I have to treat X and non-X as separate class
< ImQ009>
Thus, I need separate rows for those in the labels
< ImQ009>
And it seems I do have to add 1 to the labels, can't be 0, huh...
< ImQ009>
Any hints on that, please?
vmg17 has joined #mlpack
vmg17 has quit [Ping timeout: 260 seconds]
govg has quit [Ping timeout: 240 seconds]
govg has joined #mlpack
< rcurtin>
wow, I'll have to delete that spam from the log ...
< rcurtin>
ImQ009: I think the best example code is in recurrent_network_test.cpp, have you taken a look at that?
< ImQ009>
rcurtin, yes
< ImQ009>
That's the example I was referring to
< ImQ009>
Thing is, there are some pieces of code that example that aren't explained
< ImQ009>
Like adding 1 to the labels
< ImQ009>
rcurtin, at any rate, there's one thing that I'm very interested in getting to know: how should I go around to predicting a stream of incoming data? Can I just feed data as it comes or do I need to construct a dataset of Rho timesteps?
Addy has joined #mlpack
Addy has quit [Quit: Page closed]
vmg27 has joined #mlpack
< vmg27>
@zoq : I reinstalled boost 1.63 and the serialization issue doesnt occur anymore
vmg27 has quit [Ping timeout: 260 seconds]
dungeon_master has joined #mlpack
< dungeon_master>
Hello, I want to know if a single file can be build in mlpack? Actually I'm writing a test for a binding and I have to compile the whole mlpack or mlpack_test code which consumes a lot of time as well as computational resources.
< dungeon_master>
Moreover, I noticed that whenever I build the codebase inside Clion IDE it takes too much RAM and couldn't build ultimately but when I build the same through command line everything is build efficiently
dungeon_master has quit [Ping timeout: 260 seconds]
witness has quit [Quit: Connection closed for inactivity]