|
cheese-cube posted:figured i'd post this, Kreuzader is now bot famous and can retire it's pretty surreal to see the posts being discussed on a twitch show https://www.youtube.com/watch?v=0MDOcke676I&t=2247s but i am at the laughter it's causing
|
# ? Nov 20, 2017 17:18 |
|
|
# ? May 17, 2024 17:53 |
|
now you need to monetize
|
# ? Nov 20, 2017 19:37 |
|
would you rather: eat a nose with your couch all day or MCDONALD'S! I'M LOVIN IT!
|
# ? Nov 20, 2017 19:43 |
|
already sold out to OBJECT https://twitter.com/wyr_bot/status/931371518835216384
|
# ? Nov 20, 2017 20:31 |
|
https://twitter.com/tshirtsbot/status/932700124148654081 WEED
|
# ? Nov 20, 2017 21:05 |
|
I feel like someone did this already but has there been a SA poster boy made which has been fed GBS posts?
|
# ? Nov 20, 2017 21:26 |
|
Lewis & Quark have been baking http://aiweirdness.com/post/167736100922/a-neural-network-invents-some-pies quote:
I could go for a slice of Worf Butterscotch Pie right now
|
# ? Nov 21, 2017 20:05 |
|
It's me, the impossible maple spinach apple pie
|
# ? Nov 21, 2017 20:15 |
|
i am simply fart
|
# ? Nov 22, 2017 04:04 |
|
I'm definitely Law-Ox Strudel
|
# ? Nov 22, 2017 04:07 |
|
I'm Mur’s Tartless Tart
|
# ? Nov 22, 2017 04:10 |
|
free pie
|
# ? Nov 22, 2017 05:20 |
|
Parades Or Meat Pies Or Cake #1 checking in
|
# ? Nov 22, 2017 06:51 |
|
Dississippi Mish sounds like a rap battle move. And Scent Whoopie is definitely a euphemism.
|
# ? Nov 22, 2017 10:48 |
|
mod saas posted:Parades Or Meat Pies Or Cake #1 checking in my favorite sunset rubdown song
|
# ? Nov 22, 2017 10:58 |
|
https://twitter.com/tshirtsbot/status/929982210966544384
|
# ? Nov 22, 2017 12:59 |
|
man if that had only been a goat clipart, instant buy.
|
# ? Nov 22, 2017 13:28 |
|
i will of course be Feeb Fart
|
# ? Nov 22, 2017 14:11 |
|
https://twitter.com/KrangTNelson/status/933453238774398976
|
# ? Nov 23, 2017 06:04 |
|
|
# ? Nov 23, 2017 09:15 |
|
I have committed a tensorcrime
|
# ? Nov 23, 2017 10:05 |
|
atomicthumbs posted:I have committed a tensorcrime speaking as someone who's watched several seasons of CSI i'm p sure that's 4 low-velocity blood droplets and one big old cum stain
|
# ? Nov 23, 2017 12:22 |
|
That's, uh, an odd outlier. I'm almost afraid to ask.
|
# ? Nov 23, 2017 12:24 |
|
Collateral Damage posted:That's, uh, an odd outlier. I'm almost afraid to ask. I have assembled a bad corpus. Here's some points on PCA component axis #3.
|
# ? Nov 23, 2017 12:43 |
|
cheese-cube posted:speaking as someone who's watched several seasons of CSI i'm p sure that's 4 low-velocity blood droplets and one big old cum stain the clusters on the left are non-English languages, grouped together. The big one is definitely a cum stain, considering where I scraped this text from
|
# ? Nov 23, 2017 12:44 |
|
atomicthumbs posted:The big one is definitely a cum stain, considering where I scraped this text from new thread title spotted
|
# ? Nov 23, 2017 14:13 |
|
cheese-cube posted:new thread title spotted lol
|
# ? Nov 23, 2017 16:44 |
|
atomicthumbs posted:the clusters on the left are non-English languages, grouped together. The big one is definitely a cum stain, considering where I scraped this text from
|
# ? Nov 23, 2017 19:58 |
|
194 megabytes of mixed, unfiltered furaffinity, fed through ixa-pipes-tok for tokenization and transformation to 1-sentence-per-line format, then fed through a series of awks and seds to further format it, make lowercase, strip punctuation, etc. then fed that into gensim's word2vec script, set to skip-vector with hierarchical softmax, dimensionality of 300, and context window of 15 then feed that into gensim's word2vec2tensor script, then edit those to add a garbage line at the beginning because tensorboard projector is hosed up and otherwise will skip a line and misassociate every label. then feed that into http://projector.tensorflow.org PCA axis #5 goes from highly sexual ("pantie-clad") to non-sexual ("shogun"). PCA axis #1 seems to also have a sexual component. t-SNE shows various clusters of many, many sorts. there's a food cluster, verbalization clusters of different kinds, accented clusters, foreign clusters, and the sexual content is so pervasive it's hard to find a specific sexual cluster atomicthumbs fucked around with this message at 21:15 on Nov 23, 2017 |
# ? Nov 23, 2017 21:12 |
|
the basilisk is going to kill you very slowly for that
|
# ? Nov 23, 2017 21:29 |
|
I am now feeding the dataset into char-rnn-tensorflow, with 1 layer, 256 units, and a sequence length of 400. Using a Neural Architecture Search cell instead of RNN, GRU, or LSTM, which seems both innovative and nifty. If the output doesn't disgust me by 10 epochs I'll be disappointed. flakeloaf posted:the basilisk is going to kill you very slowly for that come at me, you bitch-rear end computer. No-cognition-having-rear end motherfucker
|
# ? Nov 23, 2017 21:33 |
|
oh no
|
# ? Nov 23, 2017 21:41 |
|
I appreciate Krang T. Nelson as much as anybody but do they really think this is a thing the world needs, or doesn't already have?
|
# ? Nov 23, 2017 23:48 |
|
Doom Mathematic posted:I appreciate Krang T. Nelson as much as anybody but do they really think this is a thing the world needs, or doesn't already have? Maybe he just wants to put those guys out of work, so they're forced to rely on the welfare state.
|
# ? Nov 23, 2017 23:57 |
|
https://twitter.com/WYR_bot/status/933727413510590465
|
# ? Nov 24, 2017 10:51 |
|
https://twitter.com/WYR_bot/status/934135162299920384 This is a tough choice, they both sound compelling
|
# ? Nov 24, 2017 21:23 |
|
https://twilightzoneintros.tumblr.com/
|
# ? Nov 24, 2017 22:10 |
|
NoneMoreNegative posted:https://twitter.com/WYR_bot/status/934135162299920384 hngh too hard to choose
|
# ? Nov 25, 2017 01:46 |
|
NoneMoreNegative posted:Lewis & Quark have been baking how did nobody claim Wrasle Cream Pie?!
|
# ? Nov 25, 2017 03:38 |
|
|
# ? May 17, 2024 17:53 |
|
https://twitter.com/WYR_bot/status/937351859693596672 if you voted the latter, there could be a career for you at the Ministry of Information!
|
# ? Dec 3, 2017 17:28 |