23
OpenAI’s new “reasoning” AI models are here: o1-preview and o1-mini
(arstechnica.com)
Haven't read the article, but I'm guessing this new model enables them to do something computers could do 20 years ago, only far, far less efficiently.
To me it seems like they added a preprocessor that can choose to tokenize letters or send directions to other systems as opposed to just using the llm. So this makes it far better at logical reasoning and math from natural language input. The achievement isn't counting the R's in strawberry, it's working around the fundamental issues of llms that make that task difficult.
death to america
I had to write some javacc parser code and javacc is sort of terrible to write so I jammed in what I wanted and o1 gave me a real and reasonable solution, as opposed to 4o which was just bad. To go a bit Ted Kaczynski, we are already seeing a huge lack of critical thinking in market decisions. I've had bad ideas thrusted upon me that were clearly chatgpt inspired, at least before chatgpt the bad ideas had a bit of novelty to them. Now they are all this homogeneous badness. Most market bets are just throw things at the wall and see what sticks and sometimes people are accidentally right, but now we have this thing where everyone is making the same bets and nothing sticks. The short term productivity gains are far out-shadowed by this larger market madness. At least now the homogeneous decisions will be a bit better now, and hopefully a bit more well reasoned, but we're still at this level of bad we can never get away from.
Another W for accelerationists? Or an L because the newer models might reproduce status quo antetreatprintus?
JavaCC
oh my god I haven't heard of that shit in YEARS lmao
https://cdn.arstechnica.net/wp-content/uploads/2024/09/strawberry_demo_HKeSJB5tjdRgdoGU.mp4?_=1
What happens in the "thinking" ?
I really need to know this.
Since you don't see any text generation happening, it could be reviewing it's own output to mimic knowledge of future tokens from a point
havent tried it much but it does appear to be much better than 4o. 4o was kinda mid, super dumb.
It's $20/mo right now for o1 and 4o is also $20/mo sans preview
this post was submitted on 13 Sep 2024
23 points (96.0% liked)
news
23576 readers
635 users here now
Welcome to c/news! Please read the Hexbear Code of Conduct and remember... we're all comrades here.
Rules:
-- PLEASE KEEP POST TITLES INFORMATIVE --
-- Overly editorialized titles, particularly if they link to opinion pieces, may get your post removed. --
-- All posts must include a link to their source. Screenshots are fine IF you include the link in the post body. --
-- If you are citing a twitter post as news please include not just the twitter.com in your links but also nitter.net (or another Nitter instance). There is also a Firefox extension that can redirect Twitter links to a Nitter instance: https://addons.mozilla.org/en-US/firefox/addon/libredirect/ or archive them as you would any other reactionary source using e.g. https://archive.today . Twitter screenshots still need to be sourced or they will be removed --
-- Mass tagging comm moderators across multiple posts like a broken markov chain bot will result in a comm ban--
-- Repeated consecutive posting of reactionary sources, fake news, misleading / outdated news, false alarms over ghoul deaths, and/or shitposts will result in a comm ban.--
-- Neglecting to use content warnings or NSFW when dealing with disturbing content will be removed until in compliance. Users who are consecutively reported due to failing to use content warnings or NSFW tags when commenting on or posting disturbing content will result in the user being banned. --
-- Using April 1st as an excuse to post fake headlines, like the resurrection of Kissinger while he is still fortunately dead, will result in the poster being thrown in the gamer gulag and be sentenced to play and beat trashy mobile games like 'Raid: Shadow Legends' in order to be rehabilitated back into general society. --
founded 4 years ago
MODERATORS