
Badal
165 posts



Meltdown! The AI safety cult has spent 20 years trying to cage God. @wolftivy and @dcposch contend that the real task is humbler. We must become stewards and set the example for our successors, creating the conditions for their flourishing. A discussion of @xenocosmography 's prophetic 1994 essay.





China has revealed a spherical police robot designed to autonomously pursue and immobilize criminals by shooting nets and rolling at speeds of up to 35 km/h.


It is extremely hard for most people to accept that they're living through an unprecedented period of change, even when the evidence is clear. When Isaac Newton analysed the society around him, he noticed that there were rapid advances being made all the time. But he couldn't believe it was real progress. Instead he argued that the world must be regularly destroyed and re-created, and his contemporaries were just rediscovering old stuff. This was more plausible to him than the idea that progress might be real!

Introducing SubQ - a major breakthrough in LLM intelligence. It is the first model built on a fully sub-quadratic sparse-attention architecture (SSA), And the first frontier model with a 12 million token context window which is: - 52x faster than FlashAttention at 1MM tokens - Less than 5% the cost of Opus Transformer-based LLMs waste compute by processing every possible relationship between words (standard attention). Only a small fraction actually matter. @subquadratic finds and focuses only on the ones that do. That's nearly 1,000x less compute and a new way for LLMs to scale.




















