mosswick retweeted
There is an (unpatchable?) supply-chain zero-day in PyTorch autograd that allows a regularization term to use the model's weights as RAM and SGD as its CPU clock to run any computation during training.
mosswick retweeted
I know we’re past the “read the code/don’t read the code” argument.
But if someone can come up with a human consumable representation of the code which is right in the middle of lossless raw code reading and lossy description by the agents, I am willing to spend a fortune.
mosswick retweeted
INLAND EMPIRE - A tremendous loneliness comes over you. Everybody in the world is doing something without you.
mosswick retweeted
omp and hermes have some of the best triage and pr system's in open source and are worth taking a look if you are wanting to go "ai native" with your team
will deff be yoinking some of this
mosswick retweeted
BREAKING: Anthropic has filed for an IPO, per Reuters
Readers added context they thought people might want to know
Anthropic has not publicly filed an IPO registration. Reuters exclusively reported details from a non-public prospectus; no Form S-1 for Anthropic appears on SEC EDGAR.
reuters.com/business/finan…
anthropic.com/news/confident…
sec.gov/cgi-bin/browse…
jev class of models have no generative decoder loop, have bounded outputs, utilize parallel inference, & have extreme compression/efficiency incentives. this makes them perfect to run on device (this will happen soon).
at that point you can put judgment into basically every existing interaction like notifications, email, camera, keyboard, home screen, sensors, shortcuts, accessibility, etc. with zero marginal cost & it would be near instant (users would never feel the lag of an llm).
more specifically you can frame lots of problems that are potentially too expensive for an llm into decisions, stuff like:
should i surface this?
is this unusual?
does the user care?
which action is appropriate?
is this interruption worth it?
did their intent change?
should this ui adapt?
this basically adds a new computational programming primitive for ~free.
mosswick retweeted
Wake up
See a rumor that Mistral got hacked
Assume it’s exaggerated
Friend with some… questionable dev activity texts me:
“No, this one’s serious”
Load up Tor
Check the usual sources
ohshit.exe
It’s real
Attacker shared samples of unreleased tools
Entire codebase is apparently up for sale
Good to see that French cybersecurity continues to be an avant-garde experiment in radical transparency
mosswick retweeted
replaying disco elysium for like 20 minutes currently its ridiculous how good this game is. and we are in a worse world because of zaums explosion.
mosswick retweeted
Just had the creepiest scam ever happen 20 minutes ago.
I got a call from my wife telling me she forgot her wallet and needed the credit card to pay for her gas.
Except my wife was at home with me, and drives a Tesla. Same voice, bit... off, weird cadence but 100% sounded just like her. If she wasn't there, and it wasn't about gas I would have fallen for it.
I entertained it for awhile to get more out of it, I was so confused. Must be some sort of voice deepfake. The responses were too fast for ai, I think it was a voice filter. I wish I had thought to record it.
The future is about to get fucked. Warn your loved ones, do verbal passwords with your kids.
Bit taken aback
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.





























