The 'Astra can read and write weights to improve Qwen' post is misleading
The experiments that read raw weights and the ones that fix Qwen use completely different techniques.
The latest blog posts to take off on Hacker News. 5,933 posts.
The experiments that read raw weights and the ones that fix Qwen use completely different techniques.
The LISP curse is often quoted as the reason why so few libraries and reusable components exist for LISP dialects. According to the concept, LISP is so power...
A fun experiment where I back-port all the improvements we've made to chess engines over the last 30 years and run it on a 1996 Pentium Pro to see what we could have done if only we knew more at the time.
How OpenAI built Pando, a Git-compatible backend designed for the scale of development that Codex makes possible.
Look at the monthly numbers...
A recent paper argues that frontier LLMs can work with neural network weights directly. This seems superhuman, but it isn’t. The paper shows that GPT-6 Astra…
The team is the first to directly observe a unique property in an altermagnetic candidate material, key evidence of a new phase of magnetism.
[This is a guest post by Karthik C. S., Amit Sahai, and Jie Xue. This blog post was initially written in a different file format and converted using AI. — T.] This is a good time for mathemat…
We create beautiful expressions through programming and deliver the best possible experiences.
No patched kernels are available yet for 22.04, 24.04, or 26.04 despite a CVSS 9.8 rating.
We find that frontier LLMs are surprisingly good at natively reading and writing neural network weights, using a variety of recent tools and algorithms or even via direct weight editing.
Wait, that's not him.
That morning I woke up at 6:58, two minutes before the alarm. I lay there and waited for it to ring. And it rang. We both did our jobs. I still don’t know why I remember it in such detail. I used to forget what I’d had for dinner the night before. And now I remember the whole morning minute by minute, as if someone…
I’m not a big owner of things, but the best laptop I’ve ever owned is a roughed-up 2015 Thinkpad X250.
Explore 125 baby names that shifted between boys and girls, from Leslie to Ashley. See how each one’s split between boys and girls changed from 1936 to 2025.
I shouldn't have to be doing any of this, but at least it's fun.
I tested thirteen models and found that their drafting is approaching professional levels.
We moved our warehouse product off DuckDB and DuckLake onto our own flavor of Trino and a new open table format, HogLake. Here's what broke at scale and why.
On reverse honeypots, rewarding close attention, and finding your people
In case you missed it: CVE-2026-98432 is a critical (CVSS 9.8) bug in the Linux kernel's TCP code. It's been there since 2016, so virtually every Linux server on the planet has it, and it can be hit remotely. Anthropic says it built a working exploit that gets root on a server. A total takeover.
Zone setting changes used to rebuild entire DNS zones, even ones with tens of millions of records. Here is how we built a planner that picks the smallest safe build, cutting full builds by 40%, peak memory by 85%, and halving build delay.
I was reminded today of AI Adoption Rate Trending Down for Large Companies, a chart produced by Apollo Global Management in September 2025, which appeared to show a slowdown in …
This story starts with a little tragedy. See, I’ve got these old webpages I love to go back and look on every now and then. There’s websites that h...
China’s First Frontier AI Export Controls: MOFCOM Announcement No. 48 Key Takeaways On October 16, 2026, China’s Ministry of Commerce (MOFCOM) and Ministry of Science and Technology (MO…
From Pascal and ex tapes to a VAX system: Bill Joy handles BSD distribution and debugging, then joins Ozalp Babaoglu to bring virtual memory to Unix.
An attacker gained unauthorized access to environment variables for a subset of organizations. We've blocked the exploit and emailed all impacted account owners with steps to mitigate.
Proton's CEO called a request for security testing "loser thinking". Using AI to write code is fine, but for a privacy company, the guardrails around that code are the product.