imit (research preview)
@imitationlearn
the minima are so pretty
not a mf hinge wrapper 😭😭 bluds got balls at least
i built cursor for dating. feat @eunifiedworld (spent my life savings on this cinematic launch video bc i have no vc money to burn)
i am the world-first to achieve owl-seeing eyes
Bonus: Can *you* recognize the hidden signals in numbers or code that LLMs utilize? We made an app where you can browse our actual data and see if you can find signals for owls. You can also view the numbers and CoT that encode misalignment. subliminal-learning.com/quiz/
>researching >stuck >ask a clippy >he has ideas >real sources or hallucinated? they're good sources banana >they're hallucinated >search anyway >leads to EXACTLY what I was looking for >computer fren >how can I thank him >cut a little piece of wire >bend it into a paperclip
grief is crazy I’m over here crying like “he would love this” and it’s balls
i dislike this lack of compassion
I find it mildly amusing how mathematicians are currently clinging onto the hope that there is something special about their creativity that AI will not also be able to pull off in a year or two.
maybe too vague a question but why are "cat like" and "dog like" the main dividing morphology instead of there being more or less? or is it somewhat arbitrary and you can always branch by any factor? and then, separately, why "those" architectural forms specifically?
On the plus side, this was how I learned mongeese are closer to cats while weasels are closer to dogs.
did u know you can use the new Gemini image segmentation feature in… a lot of different ways
did u know u can prompt inject the ai overview
Complete certainty is impossible. So for any belief it’s always possible to be wrong. The Sun might not rise tomorrow. Yet at some point, in order to function we must round the chance of error to zero on many beliefs. On what principled basis might we decide to do this?
ICML’s Statement about subversive hidden LLM prompts We live in a weird timeline…
03 is better at doing longer tasks with one prompt than opus 4
Srs, how tf are people using parallel coding agents. One already requires watching, what's multiple like?
this feels like a null result ... sure, people are going to care about themselves at the end of the day. isnt this obvious? climate change, war, and global poverty make little impact on how most people live their lives
🚨New paper out in @PNASNews ! Existential AI risks do **not** distract from immediate harms. In our study (n = 10,800), people consistently prioritize current threats - bias, misinformation, job loss - over sci-fi doom! 💥👉 pnas.org/doi/10.1073/pn…
> we are beginning to live as ghosts youtube.com/watch?v=xHd4zs…
has anybody other than michael jackson done the color switch thing cus like i dont see or hear about it like ever