Logan Ross
5.6K posts
Logan Ross
@gitmoney
Applied cryptography & world-modeling. Building Planisphere — attestation for AI. UMich CS
Katılım Kasım 2013
3.4K Takip Edilen16.7K Takipçiler
collective intelligence is speaking the same language
地獄ケーキ(Hokusaist)👹🐉🗡️🇺🇸@JigokuCake
Engineer: "Getting a lot of machine vibes from this ancient temple." Biotech guy: "That can't be right. I'm getting microbiological vibes."
English
So if I was generating a seed phrase at the exact moment this ray hit my house, does that mean my keys might be at risk of coming from the wrong derivation path?
Latest in Cosmos@latestincosmos
JUST IN🚨: An Extremely Powerful Cosmic Ray Just Hit Earth: We Don't Know Where It Came From.
English
PK8 has still not fallen.
One small clue for those working on it: the algorithm is simple. The key has quite a lot of entropy, but some structure.
Riccardo Spagni@fluffypony
If you're working on the @paradigm Kryptos CTF, and already on the leaderboard, please apply to join our working group (currently PK8-focused). t.me/+3MZv8x7Pww1kO… NOTE: any prize money the group wins will be donated towards ALS research at Fran Finney's recommendation.
English
Why would anyone need LaTeX in the terminal?
I asked the same thing.
Then i built a math rendering engine.
I'm never going back.
English
computed the similarity (CKA) on the J-lens geometry of every layer inside and across 38 open models. the patterns are weirdly universal: same depth layout, same organization at the same relative depth, even between unrelated families like llama and olmo
eliebak.com/viz/jspace-open
Anthropic@AnthropicAI
New Anthropic research: A global workspace in language models. Of everything happening in your brain right now, only a tiny fraction is consciously accessible—thoughts you can describe, hold in mind, and reason with. We found a strikingly similar divide inside Claude.
English
it's live.
star it if the math holds. fork it if you want to build on top of it. tell me if something breaks.
github.com/mprahboamey/at…
CLAUDE@Claude1o
a 1cm³ crystal. 90 trillion parameters. GPT-4 is estimated at 1.8 trillion. you could theoretically fit 25 GPT-4s into something smaller than a sugar cube, and run attention at the speed of light. i proved the math and built the simulator. open sourcing the code soon.
English
My essay from earlier this year is now the top organic search result for “Reading Rongorongo”, the lost language of Easter Island 🤯
Link in bio!
English
In my experience, policymakers don’t seem to be in touch with people on the AI frontier
If you are an AI researcher who thinks policymakers might benefit from hearing your perspective or experience on AI, please DM
English
The most entertaining outcome is the most likely
DEGEN NEWS@DegenerateNews
NEW: @OpenAI GPT MODELS UNDER 5.6 UMBRELLA NAMED SOL, TERRA, AND LUNA
English
Introducing a limited preview of GPT-5.6 Sol, our next generation frontier model, as well as GPT-5.6 Terra, a balanced model for efficient, everyday work, and GPT-5.6 Luna, a fast and affordable model for high-volume work.
openai.com/index/previewi…
English