|
too bad you can't run rnn on dead pigs
|
# ? May 4, 2016 02:20 |
|
|
# ? Jun 6, 2024 06:07 |
|
Captain Foo posted:too bad you can't run rnn on dead pigs you can if you are david cameron you can do anything you loving want to dead pigs if you are david cameron
|
# ? May 4, 2016 02:34 |
|
Sniep posted:you can if you are david cameron emphasis on loving, i guess
|
# ? May 4, 2016 15:45 |
|
Moist von Lipwig posted:it kinda fits? rofl more please
|
# ? May 4, 2016 17:07 |
|
Moist von Lipwig posted:i have no clue why it puts spaces between all the characters, this is iteration ~18000/84000 so i'm curious how it evolves from here because you're using word-rnn in character mode. try retraining with -word_level 1 and maybe -glove 1 (maybe use a bigger GloVe set than it asks for if you have time and it actually works?)
|
# ? May 4, 2016 19:26 |
|
Pumpy Dumper posted:I've got an old GTX 460. is it possible to have two different GPUs in a computer if one is just being used for CUDA? most of the rnn stuff also has an OpenCL mode.
|
# ? May 4, 2016 19:27 |
|
atomicthumbs posted:because you're using word-rnn in character mode. try retraining with -word_level 1 and maybe -glove 1 (maybe use a bigger GloVe set than it asks for if you have time and it actually works?) ahhhhh gently caress, too late now, i'll just format it code:
code:
Moist von Lipwig fucked around with this message at 22:05 on May 4, 2016 |
# ? May 4, 2016 22:02 |
|
George: It's Hell's birthday.
|
# ? May 4, 2016 22:06 |
|
Al! posted:George: It's Hell's birthday. loving goldmine
|
# ? May 4, 2016 22:23 |
|
Moist von Lipwig posted:ahhhhh gently caress, too late now, i'll just format it mail me ur corpus and i'll see what i can do
|
# ? May 4, 2016 22:54 |
|
Al! posted:George: It's Hell's birthday. I don't follow this, what's going on, is the show audio being piped into this and being voice-to-texted?
|
# ? May 5, 2016 04:10 |
|
Citizen Tayne posted:I don't follow this, what's going on, is the show audio being piped into this and being voice-to-texted? He's training it via RNN on the complete transcripts of Seinfeld. Then generating new text
|
# ? May 5, 2016 09:44 |
|
Moist von Lipwig posted:
i think we're done here
|
# ? May 5, 2016 10:01 |
|
quote:GEORGE: Hey, I love it! I'm Not George!
|
# ? May 6, 2016 19:16 |
|
goddamn please post the seinfeld corpus
|
# ? May 6, 2016 20:32 |
|
Pumpy Dumper posted:He's training it via RNN on the complete transcripts of Seinfeld. Then generating new text That owns
|
# ? May 6, 2016 20:40 |
|
AWWNAW posted:goddamn please post the seinfeld corpus http://horriblenerds.com/files/seinfeld.txt
|
# ? May 6, 2016 21:08 |
|
http://horriblenerds.com/files/i_ha..._just_scary.mp4
|
# ? May 6, 2016 21:24 |
|
no the same without the file name
|
# ? May 6, 2016 21:27 |
|
this is gold jerry!!
|
# ? May 6, 2016 23:38 |
|
8000 baby names generated by an rnn i'm Whad
|
# ? May 9, 2016 06:14 |
|
I'm either Doro-Ba, Muude, or Joaca.
|
# ? May 9, 2016 08:17 |
|
George Lucas angrily tears up a stack of papers
|
# ? May 9, 2016 08:18 |
|
atomicthumbs posted:I'm either Doro-Ba, Muude, or Joaca. I'm Aborta
|
# ? May 9, 2016 08:20 |
|
Al! posted:I'm Aborta fuckin lol
|
# ? May 9, 2016 08:34 |
|
I'm Lovandie
|
# ? May 9, 2016 13:05 |
|
i'm Furnard Hoortrane irl
|
# ? May 9, 2016 13:09 |
|
either Gritmudia or Dos
|
# ? May 9, 2016 13:35 |
|
Snapchat A Titty posted:either Gritmudia or Dos hey handsome, is there a Ms-Dos?
|
# ? May 9, 2016 13:42 |
|
|
# ? May 9, 2016 14:47 |
|
Trig Discipline posted:hey handsome, is there a Ms-Dos?
|
# ? May 9, 2016 18:06 |
|
i'm Lovynna
|
# ? May 17, 2016 23:29 |
|
i'm fatpha
|
# ? May 18, 2016 04:51 |
|
a markov style bot for thingiverse models http://www.thingiverse.com/shivinteger/about
|
# ? May 19, 2016 23:35 |
|
hey hows the seinfeld script thing going
|
# ? May 20, 2016 00:12 |
|
it's just sitting there doing nothing right now not sure what to do with it really i guess i can get it to spit out a shitload of text and put it up somewhere for people to rifle through
|
# ? May 20, 2016 00:41 |
|
bumping to note that word-rnn training on a mobile gpu (840M) is somewhat possible, but will likely need to move to EC2 soon i guess - using a 9MB corpus of TOS and TNG transcripts and all i can fit is a net size of 300 and 2 layers im encouraged by the results so far though: quote:[ Planet surface ]
|
# ? Jun 6, 2016 12:05 |
|
Crusader posted:bumping to note that word-rnn training on a mobile gpu (840M) is somewhat possible, but will likely need to move to EC2 soon i guess - using a 9MB corpus of TOS and TNG transcripts and all i can fit is a net size of 300 and 2 layers sned corpus
|
# ? Jun 6, 2016 12:34 |
|
atomicthumbs posted:sned corpus https://drive.google.com/file/d/0BxFyycj1rCP7TGFZWEtvb3RuWXM/view?usp=sharing edit - a few more quotes after additional training: quote:UHURA: I'm sorry, Mister Sulu. We have a kid. Crusader fucked around with this message at 06:13 on Jun 7, 2016 |
# ? Jun 6, 2016 13:09 |
|
|
# ? Jun 6, 2024 06:07 |
|
http://arstechnica.com/the-multiverse/2016/06/an-ai-wrote-this-movie-and-its-strangely-moving/quote:
http://video.arstechnica.com/watch/sunspring-sci-fi-short-film
|
# ? Jun 9, 2016 13:31 |