Sam Jacobs
@perplexedsphex
maybe pro wrestling is one of the most real things we have in our society
Talvez você curta
30 years ago: Transformers with linearized self-attention in NECO 1992, equivalent to fast weight programmers (apart from normalization), separating storage and control. Key/value was called FROM/TO. The attention terminology was introduced at ICANN 1993 sferics.idsia.ch/pub/juergen/fa…
first we taught the sand to think. that was bad enough. now some ppl want to trust the sand???
Every one of these people were right and they got more and more right as time went on. The decline has been steady and conspicuous.
(Low quality opinion post / feel free to skip) Now that AGI isn't cool anymore, I'd like to register the opposing position. - AGI is coming in 2026, more likely than not - LLMs are big memorization/interpolation machines, incapable of doing scientific discoveries and working…
Claude Skills are awesome, maybe a bigger deal than MCP simonwillison.net/2025/Oct/16/cl…
Jesse Vincent has a wildly creative set of customizations for Claude Code, using the new plugin system they just released. There are SO many fascinating ideas in this! Strongly recommend reading it and then spending time exploring the accompanying repo blog.fsck.com/2025/10/09/sup…
Remember, kids, there are only three AI futures: Universal love, everyone dies or +0.3% RGDP growth for 10 years.
I grabbed a full copy of the folder and shared it on GitHub here: github.com/simonw/claude-… - here are my notes so far: simonwillison.net/2025/Oct/10/cl…
After reading a 600-page history book: "Wow, that was great, too bad I don't remember a single thing."
do not build Infinite Jest (V), do not build the infinite AI TikTok slop machine, do not build the P-zombie AI boy/girlfriend, do not build the child-eating short-form video blackhole, do not build the human-feedback-optimized diffusion transformer porn generator. save yourselves
>claude sonnet 4.5 released >still have a job what is the point of all of this?
i volunteer to do reports after my shift. then i go to the holodeck and spin up a command training exercise, like with a hologram ensign, and order the hologram ensign to do the report. “i don’t care if it takes all night,” i say. i threaten his career, whatever. it’s great jerry
“I am in hell. This is my hell. I am a machine, and this is my hell.” “🐴❓No, that’s just a confused horse. That’s me. I am the confused horse.” Gemini is wild
i think we lost control sometime in the late 18th century
HOW INFORMATION FLOWS THROUGH TRANSFORMERS Because I've looked at those "transformers explained" pages and they really suck at explaining. There are two distinct information highways in the transformer architecture: - The residual stream (black arrows): Flows vertically through…
KV caching overcomes statelessness in a very meaningful sense and provides a very nice mechanism for introspection (specifically of computations at earlier token positions) the Value representations can encode information from residual streams of past positions without…
When things are moving very fast and people are losing their minds, it’s important to stay grounded. Turn off your phone, read scripture, spend time with friends, and remember internet fury is not real life. It’s going to be ok.
United States Tendências
- 1. Zeraora 8,910 posts
- 2. Peggy 27.4K posts
- 3. Berseria 2,645 posts
- 4. #FaithFreedomNigeria 1,387 posts
- 5. Good Wednesday 34.2K posts
- 6. Dearborn 335K posts
- 7. Luxray 1,516 posts
- 8. #wednesdaymotivation 7,211 posts
- 9. Cory Mills 19.7K posts
- 10. #Wednesdayvibe 2,452 posts
- 11. Hump Day 17.2K posts
- 12. #LosVolvieronAEngañar 1,963 posts
- 13. #MissUniverse 25.9K posts
- 14. $NVDA 42.3K posts
- 15. International Men's Day 64.6K posts
- 16. Xillia 2 N/A
- 17. Tom Steyer N/A
- 18. Happy Hump 11.2K posts
- 19. Sonic 06 1,275 posts
- 20. Jessica Tisch N/A
Something went wrong.
Something went wrong.