Wow.
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.…
25
Wow.
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.…
12
Super cool
Replying to @tobi
Look for Expo Mode. We built it with Google’s new HTML-in-Canvas API. I think we’re one of the first to use it.
16
Promising work and use of Claude.
Hopefully this results in new treatments.
Today, in collaboration with Anthropic, we’re excited to share rapid initial progress towards an antibody treatment for food allergy, which affects over 200 million people worldwide, discovered with the help of Claude.
At Divergence Labs, we believe that AI can unlock the next generation of antibody therapeutics with more effective mechanisms than before. Such mechanisms are urgently needed in food allergy, which involve overcoming one of the tightest affinity protein pairs in immunology, IgE and its receptor FcεRI, on mast cells.
Treatment with Xolair—the only approved antibody for food allergy—sequesters circulating IgE, but mast cells remain sensitized for weeks to months due to the extraordinarily slow dissociation of IgE from FcεRI.
We envision a treatment that actively pulls IgE off of FcεRI, without waiting for natural dissociation. Like picking a lock, this requires the right combination of geometry and force.
Our research team, working with Claude, ideated on mechanisms to accomplish this disruption, and converged on an antibody format with two arms with different functions and epitopes. Within one month, we designed, synthesized, and experimentally confirmed the biophysical function of the two arms, which we are now assembling into full therapeutic formats.
36
Very clever. Crazy AI showcase. But too nihilistic for me.
There's so much more to build.
youtube.com/watch?v=zONGVaON…
26
Love this project vision. Build it!
One of San Francisco's largest redevelopment projects could completely transform nearly 19 acres of land near Downtown.
Plans are moving forward for the redevelopment of the Caltrain Railyards at 4th and King in SoMa.
The project could include:
• Up to 5,834 apartments
• Up to 5.6 million SF of commercial space
• 160,000+ SF of retail
• A completely rebuilt Caltrain station
• More than 4 acres of public open space
• Multiple skyscrapers reaching 350 to 600 feet
• A potential 850-foot tower
This land is currently occupied by railroad tracks, platforms, maintenance facilities and the existing Caltrain station.
The proposal would modernize the transit infrastructure, relocate train storage and maintenance, and create an entirely new mixed-use neighborhood above and around the station.
Thousands of people could be living, working, shopping and spending time in what is currently a massive rail yard.
And all directly connected to one of the Bay Area's most important transit corridors.
The project is now entering environmental review, with construction potentially starting in 2029 and taking 15 to 20 years to complete....
There's obviously a long road ahead, but I love seeing developers think this big about underutilized urban land.
This could completely reshape SF.
📍San Francisco
1
51
Cool concept. Probably a lot build out here!
Can your idea become the talk of the town?
We modelled how thoughts and ideas spread in human societies by putting 2,721 simulated personas representing the US general population in an artificial city. Each persona has its own schedule, beliefs, and social circle.
Sow an idea in any persona’s mind, watch it take hold and go viral, or see it fizzle out if no one takes interest in it.
Compete with your friends and colleagues to see whose idea can win over the most adopters, get a high score and put your name on the leaderboard. Link below.
17
Super interesting. Consciousness runs at 15hz in the body.
Give vibes a whole new meaning.
Curious if the timing relates to disorders and disease. Lots of new things to discover!
1/ We discovered something totally unexpected about the neural structure of consciousness that overturns decades of thinking on its biological basis and brings us significantly closer to the recipe for generating conscious awareness.
New preprint: biorxiv.org/content/10.64898… (led by @HesseJanis, with @FrankLanfranchi, @_conor_mcgrory, @Yuelin_Shi, and @EngelTatiana)
Excited to try this! 😳
Big news for CAD: SMLIB, NVIDIA's NURBS based kernel is finally open source today 🍻🎉
We have been working with SMLIB and its dev team for more than 20 years. It enables all our products: classcad.ch headless CAD, GUI, AI/MCP, PDM, etc
github.com/NVIDIA-Omniverse/…
1
30
We have a moral obligation to oppose this.
Ro knows nothing about how to allocate $4 trillion in capital. His plan is to hand these resources to teachers unions and healthcare conglomerates.
The scale of waste and corruption is beyond comprehension.
At least FDR had a plan other than free stuff. We got bridges, parks, tunnels, and power plants from the new deal.
Prop 40 is the most important ballot initiative in America.
A 5% tax on the 989 billionaires in our country would raise $4 trillion.
We have to ask a basic question: Are we going to be New Deal Democrats or donor class Democrats?
13
Mitt retweeted
Our clinical trial of AMIE, a patient-facing AI system, was just published in The Lancet.
I want to explain what *I* think this study ACTUALLY means.
It's about a fundamental question: how do we know that clinical AI works?
🧵⬇️
nitter.cf/Google/status/21083245…
Our medical research system, AMIE, is the first patient-facing conversational diagnostic tool of its kind to be studied prospectively in a real-world clinical setting.
In a study, published today in @TheLancet, we found that when patients chatted with AMIE before an in-person appointment, the interaction built their confidence and helped them organize their thoughts. And physicians who reviewed the information from the patient conversation got more time back to focus on collaborative care and shared decision-making instead of digging through data.
Right now, I don’t think we have any good evidence to believe this.
My guess is that analog hydrocarbon systems are fundamentally different from digital metalosilicon systems.
Consciousness emerges when an information-processing system develops a sufficiently integrated, persistent model of itself as an observer
We are close 🧠
18
Great work by the Google team. Empowering patients is the future!
Today @TheLancet published our AMIE study with @BIDMC_Medicine, the first prospective study of patient-facing conversational diagnostics in a real clinic.
AMIE is a research system people can chat with before doctors appointments. Testing AMIE outside the lab with real urgent care patients showed how technology can help gather medical histories safely. Clinicians reported that the summaries helped them prepare for visits in 75% of cases and influenced their approach to care in more than half. AMIE’s differential diagnoses also matched the doctors’ final diagnoses 90% of the time giving clinicians more time to focus on patient care.
It’s Google’s first-ever paper in The Lancet’s main journal!
22
Man chases mom and 2y daughter waiting at bus stop and stabs both of them.
Girl in critical condition.
Man has prior charges for stabbing children.
Police chief, “I’ve been on parole board… justice system is working… this is just unfortunate.”
Absolutely unacceptable. We’ve let imbeciles and cowards take control of our most important institutions.
turnto10.com/news/local/prov…
40
Feel right
Codebases are going to start looking like optimized topology 3d prints
When AI is better at math than humans it also becomes better at algorithm & codebase design
At that point the system and delivering quality software doesn’t require human understanding to exist in the actual code layer.
Only needs to be understood by humans at the high level of passing testing, verification, vibe checks and being understandable when queried via AI in the rare case one need to dive deeper.
3
Really cool work.
What if,
for any scientific claim,
you can search 100M+ papers to find the strongest evidence for and against it
⚡️in 1 second?
That’s now possible with Evidence API! 🚀
paperclip.gxl.ai/evidence-ap…
For example, does drinking coffee reduce diabetes risk?
1
25
























