A man who reads old books cannot be fully captured by modern stupidity. He has dead kings, prophets, poets, killers, saints, drunks, generals, and madmen whispering in his bloodstream. The feed has no chance against this.
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.
@DistStateAndMe@willccbb bro, you quite literally shit the bed on the best thing you've been a part of and will ever be a part of. You are the last person to give any advice on here whether it be satirical or serious in nature.
@alex_dreyfus@zondacrypto It's true. It's currently a big news item in PL. For those that don't remember, Zonda is the rebrand of Bitbay. Bitbay also had it's fair share of... "questionable" headlines in its day.