irfan retweeted
We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math.
It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training.
We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵
Blog: rikiyat.github.io/blog/weigh…
Readers added context they thought people might want to know
LLM was given weights explicitly as text in prompt and used code/forward passes (per authors' logs) to ablate Qwen looping direction, a known thinking-mode issue per model card. Primarily fixed repetition in that mode, not general 'natively' improving weights/math.
rikiyat.github.io/blog/weight-sp…
github.com/RikiyaT/lm_und…
huggingface.co/Qwen/Qwen3.5-0…
huggingface.co/RikiyaT/Qwen3.…
irfan retweeted
The thing I learned this year is that when engineers say they like working on “hard problems”, they actually don't. What they like is puzzle solving, and having a repertoire of patterns that they can pattern match onto traditionally hard problems, usually involving databases or something with scale, etc. And they know that the industry used to pay you a lot for how deep and broad this repertoire was. Unfortunately, in 2026, well, you have a cheat code machine that has infinite breadth and depth for pattern matching. So the question is, do you actually like hard problems? Because there are now hard problems that are a complection of engineering, product, society, culture, politics, and they are fairly intractable, probably unsolvable! Because if you do want to work on those actually hard problems, there is so much money that people will pay you for it. But you’ll have to face that truth about yourself first.
irfan retweeted
Cool eval. Simply ask an LLM “Land or Water?” and give it a latitude and longitude coordinate as text. Ask 16,200 times, plot as image. The models know. From compressing the internet.
irfan retweeted
There’s a joke that goes something like “animals are a technology invented by plants to move seeds around” and this is the sense in which I think AI is a technology invented by humans to perform cognitive tasks.
irfan retweeted
I overwrote my keyboard's firmware so that it runs a tiny character-level language model to predict the next key you'll hit. It has per-key backlight LEDs and just lights up the ones you're likely to hit. All inference local on the microprocessor - works without the computer too!
irfan retweeted
new on diiverge: turn any photo into an explorable world ✨
diiverge.co/studio
click anything in it, pick what happens, watch it play.
share the link so friends can walk your paths, or open it and let them carve their own.
Aaaaaaand we're live!
diiverge.co
A persistent, infinite, point-and-click adventure.
Every picture is a fork. Click something in it, decide what happens, and the world grows in that direction.
irfan retweeted
Did you know your car's Tire Pressure Monitoring System transmits uniquely identifiable information whenever you're driving? I hooked up a radio dongle to my computer and monitored the neighborhood for an evening. This is what Claude inferred from the data.
irfan retweeted
Virtually no one who cites a need to "beat China" in the "AI race" ever bothers to explain what any of that actually means: what the stakes are, where the finish line is, why we're racing in the first place, what "winning" looks like, etc. Largely vibes-based techno-nationalism.
irfan retweeted
the naming of companies, as elon says, is doomed to anti-nominative determinism: openai is not open, and anthropic will lead to the end of the anthropocene. and the same: spacex is doomed to geocentrism, cannabalized by earthly interests (compute), not exploration of the stars
irfan retweeted
Meet DiffusionGemma!
An experimental open model that explores a fast approach to text generation, released under an Apache 2.0 license.
Moving beyond sequential, token-by-token processes to generate entire blocks of text simultaneously. Here’s what’s new with DiffusionGemma: 👇
