|
https://twitter.com/feelingoldbot/status/1096017466269020161 e: gently caress https://twitter.com/feelingoldbot/status/1093556249516494850 Carthag Tuek fucked around with this message at 18:48 on Feb 14, 2019 |
#
?
Feb 14, 2019 18:44
|
|
|
|
| # ? Jun 5, 2024 20:24 |
|
|
Krankenstyle posted:https://twitter.com/feelingoldbot/status/1096017466269020161 i was born closer to the korean war than to today
|
|
#
?
Feb 14, 2019 22:00
|
|
|
https://github.com/openai/gpt-2 input: "lol", flags: --top_k 40 --temperature 0.7 --batch_size 3 --nsamples 3 output: quote:======================================== SAMPLE 3 ======================================== Yeah.
|
|
#
?
Feb 15, 2019 07:34
|
|
|
Trabisnikof posted:https://github.com/openai/gpt-2 Wow, AI independently developed TVTropes
|
|
#
?
Feb 16, 2019 03:52
|
|
|
I've enjoyed this thread for a couple years now so I finally figured I should put my CS degree to use outside of work and make a dumb text bot. Wikipedia has lists for everything these days and sometimes they're really specific. There's actually about 100,000 'List of' articles on Wikipedia so I used those as my source data. Now I can generate future wikipedia list articles! https://twitter.com/WikipediaListOf https://twitter.com/WikipediaListOf/status/1094414782088663041 https://twitter.com/WikipediaListOf/status/1096685270533980160 https://twitter.com/WikipediaListOf/status/1096685828154089472 https://twitter.com/WikipediaListOf/status/1096885080398606336 There's definitely a lot of hot garbage being generated but I'm mostly just happy it works but I can make the generated titles a lot better with some tweaking over time. Thanks for the inspiration yospos
|
|
#
?
Feb 16, 2019 23:16
|
|
|
https://twitter.com/WikipediaListOf/status/1094415383602200576 https://twitter.com/WikipediaListOf/status/1094413458114637825
|
|
#
?
Feb 16, 2019 23:26
|
|
|
lol those are all great.
|
|
#
?
Feb 16, 2019 23:38
|
|
|
stifling irl lols, good work
|
|
#
?
Feb 17, 2019 06:27
|
|
|
https://twitter.com/WikipediaListOf/status/1094414092859658240
|
|
#
?
Feb 17, 2019 06:28
|
|
|
also itll probably be welcome in the pyf procedural content thread https://forums.somethingawful.com/showthread.php?threadid=3843860
|
|
#
?
Feb 17, 2019 07:05
|
|
|
https://twitter.com/WikipediaListOf/status/1096904343251431425 https://twitter.com/WikipediaListOf/status/1096820155244371969 my girlfriend is Legos fucked around with this message at 08:50 on Feb 17, 2019 |
|
#
?
Feb 17, 2019 08:48
|
|
|
this one's real
|
|
#
?
Feb 17, 2019 11:52
|
|
|
Yesterday at lunch my dad speculated that drug names were procedurally generated (he didn't exactly have the vocabulary to express it in quite this way, he's just a retired lawyer who used to read PC Magazine in the early 90s). I'm wondering if he's right though?
|
|
#
?
Feb 17, 2019 14:28
|
|
|
The brand names or the drug names themselves? The drug names go according to scientific conventions, and the brand names have to follow an international convention for pronunciation and spelling and are extensively market tested. While drugs are in trials, they are usually called two letter - three number names, like RU-486, which lay people only tend to know about if something goes sideways in testing. We test a Lot of drugs that don't make it to market so trialed drugs don't get product names until pretty late in the process. The product names also affect the placebo and nocebo effects. For example, you might think about bupropion, which is called Wellbutrin when marketed for depression (or off label in the US for ADHD but on label in Canada for the same purpose. But called Zyban when used as a smoking cessation tool. I don't know why Well- seems to make the anti-depression placebo stronger (please ignore the fact that "anti-depression placebo" may be literally non-sensical for the purposes of this unrelated point) or -ban makes the smoking cessation effect stronger, but I'm sure it does because millions of dollars go into R & D for drug marketing. I mean maybe the brand names they trial are procedurally generated though?! WrenP-Complete fucked around with this message at 15:49 on Feb 17, 2019 |
|
#
?
Feb 17, 2019 15:46
|
|
|
Schadenboner posted:Yesterday at lunch my dad speculated that drug names were procedurally generated (he didn't exactly have the vocabulary to express it in quite this way, he's just a retired lawyer who used to read PC Magazine in the early 90s). I'm wondering if he's right though? you need a third computer that makes sure each generated name doesn't mean something offensive in another language
|
|
#
?
Feb 17, 2019 15:50
|
|
|
Schadenboner posted:Yesterday at lunch my dad speculated that drug names were procedurally generated (he didn't exactly have the vocabulary to express it in quite this way, he's just a retired lawyer who used to read PC Magazine in the early 90s). I'm wondering if he's right though? The manufacturer names are pure branding but the generic names of certain kinds are based on formulas. For example https://en.m.wikipedia.org/wiki/Nomenclature_of_monoclonal_antibodies
|
|
#
?
Feb 17, 2019 16:29
|
|
|
WrenP-Complete posted:The brand names or the drug names themselves? The drug names go according to scientific conventions, and the brand names have to follow an international convention for pronunciation and spelling and are extensively market tested. While drugs are in trials, they are usually called two letter - three number names, like RU-486, which lay people only tend to know about if something goes sideways in testing. We test a Lot of drugs that don't make it to market so trialed drugs don't get product names until pretty late in the process. The product names also affect the placebo and nocebo effects. Yeah, the marketing department goes through dozens of names before settling on the one they deem the most effective. For example, the runner-up name for Viagra was Mycoxafloppin
|
|
#
?
Feb 17, 2019 18:41
|
|
|
snort
|
|
#
?
Feb 17, 2019 23:01
|
|
|
Sagebrush posted:Yeah, the marketing department goes through dozens of names before settling on the one they deem the most effective. For example, the runner-up name for Viagra was Mycoxafloppin IDGI? 
|
|
#
?
Feb 18, 2019 00:11
|
|
|
WrenP-Complete posted:I mean maybe the brand names they trial are procedurally generated though?! After enough market research and feedback I think you’d be hard pressed to tell the difference between that and the sales department.
|
|
#
?
Feb 18, 2019 00:19
|
|
|
Schadenboner posted:IDGI?  My cock’s a flopping.
|
|
#
?
Feb 18, 2019 00:40
|
|
|
tmi dude, tmi
|
|
#
?
Feb 18, 2019 00:42
|
|
|
https://twitter.com/WYR_bot/status/1097401959739547650 "Just kill me" answer 33% of users.
|
|
#
?
Feb 18, 2019 10:10
|
|
|
death is tempting but otoh my future self can hardly be worse than my present self
|
|
#
?
Feb 18, 2019 10:20
|
|
|
it's finally made some coherent ones https://twitter.com/WikipediaListOf/status/1097497940443029504 https://twitter.com/WikipediaListOf/status/1097399735848075264 https://twitter.com/WikipediaListOf/status/1097543263731347456
|
|
#
?
Feb 18, 2019 18:56
|
|
|
I thought the Sputnik one was a real article?
|
|
#
?
Feb 18, 2019 19:59
|
|
|
https://twitter.com/EndlessJeopardy/status/1097288432236154880
|
|
#
?
Feb 18, 2019 21:11
|
|
|
the replies 
|
|
#
?
Feb 18, 2019 22:52
|
|
|
just got “Merry Christmas and welcome to my next post.” out of gpt-2 and i think im done here
|
|
#
?
Feb 20, 2019 05:54
|
|
|
https://www.thiswaifudoesnotexist.net/
|
|
#
?
Feb 26, 2019 19:14
|
|
|
Hey up in Canada we have a grant called IPF for web series. We made a trailer for our series about AI or machine learning being used to replace copywriters and editors and that AI running buckwild and taking over and trying to replace everybody. If we can prove there is an audience for it, the grant is pretty huge (we are looking for $150,000 which is the low end of what people ask for) and we could make a whole series of janked out computer videos of an algorithm taking over. Give it a watch. Maybe you will like it, cause if you are here you like both this bullshit comedy and computers trying to "learn". Or like hate it. Honestly I think our pitch is good, this trailer is... fine. https://www.youtube.com/watch?v=6B809gzoo-U
|
|
#
?
Mar 5, 2019 08:32
|
|
|
https://twitter.com/MagicRealismBot/status/1106459905152577536
|
|
#
?
Mar 15, 2019 09:41
|
|
|
jeez youd think he could spare one for lowtax
|
|
#
?
Mar 15, 2019 09:58
|
|
|
Krankenstyle posted:jeez youd think he could spare one for lowtax where do you think he got them all from?
|
|
#
?
Mar 15, 2019 11:27
|
|
|
https://twitter.com/WYR_bot/status/1106492821626855424?s=19 wyr bot is this about the nfl
|
|
#
?
Mar 15, 2019 14:26
|
|
|
my girlfriend is Legos posted:https://twitter.com/WYR_bot/status/1106492821626855424?s=19 correct one is winning too 
|
|
#
?
Mar 15, 2019 14:34
|
|
|
Trig Discipline posted:where do you think he got them all from?  lowtax had 360 spines?!?1+
|
|
#
?
Mar 15, 2019 14:35
|
|
|
Krankenstyle posted:
he thought he couldn't lose. his hubris was his downfall
|
|
#
?
Mar 15, 2019 14:41
|
|
|
its called 360 spines because you spin 360 degrees and collapse into a soft heap
|
|
#
?
Mar 15, 2019 14:47
|
|
|
|
| # ? Jun 5, 2024 20:24 |
|
|
Krankenstyle posted:its called 360 spines because you spin 360 degrees and collapse into a soft heap
|
|
#
?
Mar 15, 2019 15:29
|
|