00:01
<
rcurtin >
alsc: the basic idea looks good to me, we can definitely work that into a PR and merge it
05:26
vivekp has quit [Ping timeout: 264 seconds]
05:27
vivekp has joined #mlpack
05:29
kaushik_ has joined #mlpack
06:40
jas__ has quit [Ping timeout: 256 seconds]
06:53
jas__ has joined #mlpack
07:15
alsc has joined #mlpack
07:42
alsc has quit [Quit: alsc]
07:46
kaushik_ has quit [Quit: Connection closed for inactivity]
08:09
jas__ has quit [Ping timeout: 248 seconds]
08:19
alsc has joined #mlpack
08:21
<
alsc >
zoq: it's a bit blurry how to input labels to FNN:Train
08:22
<
alsc >
I'd expect a one-hot vector, but it's not like that. looks like it's 1-based position in a vector of unique-labels?
08:23
<
alsc >
while the results matrix is 0-based.. I don't get it
10:03
alsc has quit [Quit: alsc]
10:26
deepanshu_ has joined #mlpack
11:08
alsc has joined #mlpack
11:09
alsc has quit [Client Quit]
11:24
alsc has joined #mlpack
11:50
<
zoq >
alsc: Ah right, the NegativeLogLikelihood is 1 based, I'll adjust that later today.
11:51
<
alsc >
I have checked the source code yeah, got it. why not accepting one-hot?
11:53
<
zoq >
the data I used for testing was just like that ...
11:58
alsc has quit [Quit: alsc]
12:41
jas__ has joined #mlpack
12:41
jas__ is now known as preet
13:06
preet has quit [Ping timeout: 248 seconds]
13:07
preet has joined #mlpack
17:56
preet has quit [Ping timeout: 256 seconds]
18:45
rits has joined #mlpack
18:45
rits has quit [Client Quit]
18:46
rits has joined #mlpack
18:47
rits has quit [Client Quit]
22:55
deepanshu_ has quit [Quit: Connection closed for inactivity]
23:08
dany has joined #mlpack
23:09
<
dany >
rcurtin: hey! I do check the logs, saw the pull request
23:10
dany has quit [Client Quit]
23:10
dany has joined #mlpack
23:10
dany has quit [Client Quit]
23:36
<
rcurtin >
dany: great, hopefully it should fix your issue---if not (or if you find more) just let me know