This is absolute nuts , RSI is here xd!
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…
Jitesh Nayak 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…
Jitesh Nayak retweeted
Weekend hack: HackerTube. I was annoyed with YouTube's recommendations so I had Claude filter down 43M comments over 20 years of HN to 6,454 videos that keep being recommended. *Significantly* better viewing experience, ~2 hours with Opus 5.5:
hackertube.org/
- Excellent recs without any logins (state is kept in local browser session)
- Thumbnail is a representative frame from middle of the video instead of the super duper A/B test hyper-optimized user uploads
- Underlying data from the comment analysis as one zip file, point your Claude at it and ask for your own version
- HackerTube Longs ✊
A bit more on the process:
I use Claude Code (desktop app for Mac), Opus 5.5 on Medium effort as my default. I only talk to Claude and have it do everything, the only "manual work" I helped with in tandem was in the browser (reminiscent of MenuGen): 1) I had to create some kind of API key from Google Cloud UI to query YouTube for video metadata, 2) I booted up a VPS box for serving and 3) I bought the domain name. Once the box was up I gave Claude the IP and told it update, secure, harden, deploy and optimize the code, and it did the whole thing.
Many awesome projects that would have normally taken weeks/months even can now be few-shotted for breakfast. What else are we not thinking of.
Jitesh Nayak retweeted
In my view, the right response by the mathematical community to OpenAI’s papers is to say:
“We’re not accepting these papers as finished work. It’s not our job to sit down and read your model-generated proofs.
It is your job to present your work through presentations and convince people of the correctness of your work. It’s always the case that whoever makes a claim needs to provide sufficient explanation and evidence to make independent verification feasible.
Until that moment, you just have a claim.”
There’s a difference between a proof being mathematically correct and a proof being accepted as correct by the mathematical community. Mathematical truth doesn't depend on consensus, but scientific recognition requires scrutiny, communication, and reproducibility.
They're still the arbiters of whether anyone's claims, including OpenAI's, are accepted as valid, and they should leverage that to the maximum extent possible.
Jitesh Nayak retweeted
Terence Tao made a very powerful point:
“In most cases in pure mathematics, the problems are posed not because we desperately want the solution to these problems in and of themselves, but because we have seen from past experience that human-directed efforts to solve these problems tend to spur further development of the field through the efforts to solve such problems, and then to digest any partial or complete solutions that emerge for further insights.
Prematurely solving the problem by purely AI-powered methods—particularly without full transparency into the solution process—can contaminate this process to the point where it actually becomes a net negative for the progress of mathematics as a whole.”
I’ve been following Terence Tao for like 20 years, and I have always been impressed by the clarity of his thinking.
This clarity stands out even more in this difficult moment, as AI labs turn mathematics into a battleground in their race for supremacy.
In this race, they risk damaging one of humanity’s highest intellectual endeavours.
Jitesh Nayak retweeted
I've got an agent in a loop optimizing a renderer with the goal to minimize frame times (and tests to measure). It got times down from 88ms to 2ms and allocations down from ~150K to 500. Sounds good, right? Wrong. This is exactly why agent psychosis is a big fucking problem.
As an experiment, I rewrote the Ghostty core render state in Go, with access to identically laid out data structures as Ghostty and the exact same validation tests. I made a purposely naive renderer (simple, correct, but slow). 88ms per frame with 150,000 allocations (horrendous, lol)!
I then kickstarted a Ralph loop to bring the frame times down. I told it it can't modify input data structures or the public API or tests (they're correct), but it can do anything else it wants. It got to work.
It has worked for about 4 hours. I've spent around $350 on this experiment so far. The results?
88ms => 1.5ms
150K allocs => ~500 allocs
Incredible right? Nope.
My hand-written renderer I ported has frame times (same benchmark) of ~20us (0.020ms) and 0 allocations in the update path.
This is the problem with psychosis and lacking systems understanding. If you don't understand the system, you're going to accept that this is an incredible result. If you understand the system, you'll see better solutions immediately and can do roughly 75x better on throughput.
The people who blindly trust agent output are in the former camp. They're sheeple, overdrinking from a fountain of mediocrity.
Standard disclaimer: I use AI all the time. I like AI. The point I'm making is to not blindly accept results. Think. Analyze. Learn.
Jitesh Nayak retweeted
Three minutes of me driving around the GTs, passing P2s, spinning @George_Kurtz, and going three wide at ~300 km/h in turn 1 at the Daytona Rolex 24 with the Shopify #11 LMP2. Pure flow at the edge of adhesion!
Jitesh Nayak retweeted
We are a Small Research Lab based out of india
and we just dropped a one of a kind SoTA multimodal multilingual document retrieval model.
can embed and query documents across 22 langauges :
English, Spanish, French, German, Italian, Hindi, Marathi, Sanskrit, Kannada, Telugu, Tamil, Malayalam, Chinese, Japanese, Korean, Arabic, Bengali, Gujarati, Odia, Punjabi, Russian, and Thai.
It achieves state of the art performance across cross lingual and cross modal tasks.
You can now query a corpus of Japanese documents in Hindi or any of the other language.
check it out 👇
demon slayer infinity castle, was the best animated movie i watched period , visual treat
#DemonSlayerInfinityCastle
Jitesh Nayak retweeted
Sorry you don't like our codec implementations in VLC. Looking forward to seeing yours.
This quoted post is unavailable.
Jitesh Nayak retweeted
I interned at an IIT and built the entire codebase for a research project. Just saw the paper published in a journal, with zero mention of me as an author or contributor.
I did the work. They took the credit. This is beyond unfair.
:D
Doomed!
We just saw an AI design a millimeter-wave power amplifier from scratch.
Not optimize it—design it.
→ It chose the architecture.
→ Sized the transistors.
→ Generated the matching networks.
The result?
A working PA that looks nothing like human design.
The matching networks resemble QR codes more than spirals.
This was presented at ISSCC 2025 by Princeton
→ Ten days of training on 400 CPU cores.
→ Minutes to get a ready-to-fab circuit.
The hardware design paradigm is slowly evolving:
→ From design tools to design agents.
→ From writing netlists to giving design intent.
→ From tuning circuits to describing goals.
This is what “vibe design” could look like in hardware.
A new workflow where the engineer sets the direction.
And the machine does the rest.