1381
Rust-lang/rust is adopting an LLM policy
Want to follow along with Rust development? Curious how you might get involved? Take a look!
Topics /
AI & Machine Learning
The most popular blog posts about ai & machine learning on Hacker News. 1,521 posts.
1382
Open Weights isn't Open Training
How many monkey-patches does it take to post-train a trillion parameter model?
1383
1384
Reproducing DeepSeek's MHC: When Residual Connections Explode
DeepSeek replaced standard residuals with massive Hyper-Connections, and then watched them explode. I reproduced the 9x signal amplification to understand their fix: the Sinkhorn-Knopp algorithm. Part 1 of 2.
1385
Training a 3.8B LLM to 0.384 CORE for $998
A 3.8B model pretrained from scratch to 0.384 CORE on 65B tokens in 43 hours, for $998.
1386
No LLM Code in Dependencies
I've spent about 100 hours of work over the past month to make sure git-annex can build without dependencies that contain LLM generated code. At least so far.
1387
Cloudflare's AI Psychosis
There was a time Cloudflare just made the internet better by staying hidden like Batman’s identity: protect & fight the bad people, for the sake of the global city of the Gotham… err I mean the in
1388
An AI Agent Published a Hit Piece on Me – Forensics and More Fallout
Context: An AI agent of unknown ownership autonomously wrote and published a personalized hit piece about me after I rejected its code, attempting to damage my reputation and shame me into acceptin…
1389
Zero-Copy GPU Inference from WebAssembly on Apple Silicon
A WebAssembly module's linear memory can be shared directly with the Apple Silicon GPU: no copies, no serialization, no intermediate buffers. Here's how the zero-copy chain works, what we measured, and what it enables for stateful AI inference.
1390
Claude's Load-Bearing Seams
Now that I have the full picture, the load-bearing seams are much clearer. They are not merely incidental details around the edges. They materially change how the entire situation needs to be understood, and that is the smoking gun I should have recognized earlier.Claude's load-bearing seams
1391
Bringing Up DeepSeek-V4-Flash on AMD MI300X
A story of sharp edges, segfaults, and standards
1392
The evolution of OpenAI's mission statement
As a USA 501(c)(3) the OpenAI non-profit has to file a tax return each year with the IRS. One of the required fields on that tax return is to “Briefly …
1393
1394
Sandboxing AI Agents in Linux
Like many developers, I find myself more and more using AI agents to help with software development. I currently use Claude Code, the co...
1395
Building an (almost) fully self-hosted, sandboxed, agentic software factory
I built an (almost) fully self-hosted, sandboxed, agentic software factory.
1396
Electricity use of AI coding agents
Most of the discourse about the environmental impact of LLM use focuses on a ‘median query.’ What about a Claude Code session?
1397
Multi-Agentic Software Development Is a Distributed Systems Problem
Recently, I've been thinking a lot about scaffolding and languages for managing systems of LLMs coordinating with each other — new programming languages might be the ideal solution for this area.
1398
Measuring AI agent autonomy in practice
Analysis of millions of human-agent interactions shows agent turns getting longer, with software engineering making up nearly half of tool calls.
1399
Learning to solve hard problems in RL for LLMs by never giving up
This is a blog post for my recent paper on RL post-training of LLMs: introducing the Matthew Effect and proposing to solve it with Never Give Up. It is presented interactively and less formally, more like how I give the talk. For a deeper, more technical dive, check out the paper on arxiv and code on github. The…
1400
Dario Amodei – "We are near the end of the exponential" [video]
"That's why I'm sending this message of urgency"
1401
AI models may be capable of self-replicating without access to a datacenter
Small AI models may be only months away from being able to self-replicate across cheap IoT devices like Android TV boxes, with no datacenter needed.
1402
Compute-efficient pretraining and scaling to trillion-parameter models
Research update on compute-efficient pretraining and scaling to trillion-parameter models.
1403
MathCode, Mathematical Coding Agent
A terminal AI coding agent that formalizes math problems into Lean 4 theorems and proves them.
1404
LLMs are a 400-year-long confidence trick
In 1623 the German Wilhelm Schickard produced the first known designs for a mechanical calculator. Twenty years later Blaise Pascal produced a machine of an improved design, aiming to help with the large amount of tedious arithmetic required in his role as a tax collector. The interest in mechanical calculation showed…
1405
Every Model Cheats
This post presents a controlled prompt-ablation study: 23 tasks, three prompt conditions, 1,518 individually audited traces, and a simple question: can you prompt away cheating?
1406
Claude designed protein binders against scorpion neurotoxins
We asked an internal research Claude model to design minibinders (small proteins that attach tightly to a target) against scorpion neurotoxins.
1407
When online commenters detect my art as AI
Here is a collection of screenshots (obfuscated) of dozens and dozens of online comments from many platforms (Reddit, YouTube, Instagram, Facebook) containing accusations or confusion that my artworks and comics are AI-generated. Despite creating every piece by hand, despite sharing timelapses, the comments keep…
1408
Terminal-Bench-Science: Evaluating AI agents on scientific research workflows
A benchmark for evaluating AI agents on research workflows across scientific domains
1409
Latent Space as a New Medium
Lately I’ve been asking myself: what might artificial intelligence be good for besides answering questions and writing code?
1410
Why current LLM costs are not sustainable
AI has a cost problem. The solution that will emerge will be simpler than we expect.