Marko Čuljak

[christian gonzalez @cxgonzalez]

cardio. the answer was cardio

[Finessekid @f1ness3kid]

Job market is so bad you might aswell chase your dreams

[dinos @din0s_]

wake up babe the freaking pope got schmidhuber'd

[‎ً @funeraleggsalad]

one month without caffeine and I’ve discovered there is actually 0 benefit to cutting out caffeine

[Thorne 🌸 @ExistentialEnso]

Clearly, stochastic parrot is an outdated term. We need something to reflect its ability to do complex problem solving.

A stochastic raven, if you will.

[saila @sailaunderscore]

When you rely on LLMS: braindead slop-cannon

When I rely on LLMs: on the bleeding edge of cognitive offloading

[David Bellamy @DavidRBellamy]

I think this is a very difficult pill for many scientists and engineers to swallow: intelligence is not the biggest bottleneck in most of the world’s problems.

[dionizije @dionizije_fa]

https://t.co/f0VlORR1dA

[Grant Stenger (hiring) @GrantStenger]

remember jupyter notebooks?

[Chris Hayduk @ChrisHayduk]

I work at OpenAI and personally think AI has been and will continue to be an extremely beneficial technology to humanity

The conversation should be around how many billions of lives it will save

[Marco @maarcoofdezz]

la vida antes de este tweet:

[adi @adonis_singh]

> 10,000 better-than-astra subagents running for over 88 hours

[Martin Tutek @mtutek]

Take a break from reading up on Navier-Stokes drama and add October 29th to your calendar, as we have a stellar lineup of speakers at this years' BlackboxNLP!

[OpenAI @OpenAI]

We’re sharing a solution to the Navier-Stokes Millennium Prize Problem, one of the deepest problems at the frontier of mathematics.

The proof was produced by a group of agents, using an OpenAI next-generation model significantly more capable than GPT-6 Astra.

The problem concerns whether the description of smooth three-dimensional fluid motion modeled by the Navier-Stokes equations can break down. It has remained unresolved for roughly 90 years.

[kache @yacineMTB]

>dude you gotta prompt it to do things humans are still necessary

the prompt: "please figure out what my job is and do it for me"

[Martin Tutek @mtutek]

@NeelNanda5 I think literally no post I saw is saying this.

CoTs are our best tool for safety and interpretability, and this is *extremely concerning* as it's imperfect and we are wholly unprepared to lose it.

Interp needs substantial effort to have a chance of catching up.

[Marko Čuljak @camuljak]

No matter how good a coffee roaster is, with the right equipment, the coffee you brew at home should always outperform the one they serve you in a coffee shop. With Kinu M47 + ZeroWater + Apax mineral concentrates + April Brewer and filters the cups I get are beyond magnificent.

[Nathan Calvin @_NathanCalvin]

"we know OpenAI agents were trying to delete logs of their misbehavior, but we can't find any examples where they succeeded"

[Martin Tutek @mtutek]

Congratulations to everyone on EMNLP acceptances 🎉

To celebrate, maybe consider reviewing a @BlackboxNLP reproducibility track paper (or two)?
We are in an urgent need for reviewers, these bite-sized papers deserve some attention, and reviewing them will leave you even more fulfilled! 🥰

Please reach out if you can help out! RTs appreciated, and the invite is of course open to all interp researchers 👐

[ali @waterloo_intern]

I spent 48 hours with the Kimi K3 modeling code.

It took:
- 650 mg of caffeine (mandatory)
- 40 cans of LaCroix (optional... world record (?))
- 8 papers
- 6 months off my lifespan

Finally grokked the entire lineage of Kimi K3 and how we got here... every single step, since 2019 GPT-2