aliens could land tmr and it would be the 4th most surprising thing that's happened recently
All 5 genetic “letters” of DNA and RNA discovered in a 4.5-billion-year-old asteroid sciencedaily.com/releases/20…
I found 12 new ISS photos of Red Sprites (on top of the 35 already tagged by NASA) by scanning 1.9 million untagged ISS night photos.
Red sprites are beautiful electric discharges occurring ~50-90 km above thunderstorms, and last for barely a frame.
Ground photo: Paul M. Smith
All 12:
ISS043-E-262974
ISS053-E-162720, 236710, 264396, 288056, 361847
ISS055-E-35007, 47451, 91395
ISS068-E-1270
ISS074-E-774567, 780091
Images: NASA Johnson Space Center (eol.jsc.nasa.gov)
They thought he was speaking gibberish but he was just Hungarian 😭
Looking at the reasoning trace, astra identified the games by:
1) which of the 10x10 grid cells had strong weights
2) which actions had large biases
E.g.
- Pong: only up and down actions have strong biases
- Breakout: bottom row for paddle, bricks at top
- Freeway: strong bias towards moving upwards
- Pac-Man: maze is visible from grid input weights
Still cool, but probably a game enthusiast would have figured it out in a few minutes in a similar way.
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.…
This made me check whether today was April 1st.
I can see how this could be done with a lot of reasoning.
If Astra could, on zero-thinking mode, identify the Atari game based on weights or write functional rl policies, this would shatter my world view on the strength of LLMs.
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.…
First time using muse today - to make Metrograph accounts and purchase tickets:
- turns took ~15+ min and got stuck without alert
- muse can't reliably hold 2 logins for one site
For science, I kept trying. After ~4.5 hours, I gave up and only 1/2 was purchased. @alexandr_wang
ryan retweeted
If you want to escape the insanity you can also come to Prague.
- Build foundation models. Lots of compute.
- Work on hard fun unsolved problems.
- We're already very profitable.
- Idea -> prod in a week.
- No sales. Ever.
- Small team.
- RL.
👋
The people of Silicon Valley are unwell.
If you work at a frontier lab or decacorn, you’ve got it bad. You haven’t seen your kids, your girlfriend’s annoyed, invites from friends to hang out are waning, even weddings are on hold. This is the time to make your mark and make your money, after all.
If you’re building a startup that’s trying to compete with or be one of these companies, you’ve got it bad. Maybe you passed up a job offer at one a couple years back, which means you left $100M on the table, you fucking idiot. Being an underdog isn't as fun as you thought.
If you’re working on a pre-AI startup, you’ve got it bad. Not too long ago you were the belle of the ball, but now your 30% year-on-year growth doesn’t feel like growth at all. It feels like death.
"Too many of my friends are spiraling," Brie Wolfson writes. "I don’t know how we’re supposed to live and work like this."
It’s all too much, too fast. Too much money, too much growth, too much power, too much change. Every month, a new frontier model is released. Every day, a new unicorn is minted. Seriously. 300 days into the year 2026, and 311 unicorns have been minted so far.
So what’s left for the sane ones? Brie went looking for teams playing a different game. She found one in Ottawa: a search infrastructure company that many of the AI darlings rely on.
It's growing fast, and its people are well.
Read @zebriez’s latest company dispatch from inside the sane asylum of turbopuffer.
colossus.com/article/sane-as…
ryan retweeted
Replying to @small_LM
I don't think the world can handle the level of randomness emanating from this account.
thank you for the endorsement @AgustinLebron3
Replying to @AgustinLebron3 @small_LM
A million monkeys and a million typewriters?
Nope. Just Ryan on a rainy Sunday.
She’s a MIGHAgEAMBMGByqGSM49AgEGCCqGSM49AwEHBG0wawIBAQQgrx+cHILcssolsRxH0bihENKkin9Bvr06n5vhlzCC7XuhRANCAARCp18iaChXwQ3YvdIMmCJF8D8dj9q7BaU1fKfJIxM7UaZXXX4ATAztgD9Xt1avFvpHFn8ZIVa0tM71X8+P93io



















