How to Make a Fast Dynamic Language Interpreter
This post is about optimizing an extremely simple AST-walking interpreter for a dynamic language called Zef that I created for fun to the point where it is competitive with the likes of Lua, QuickJS, and CPython. Why? Most of what gets written about making language implementations fast focuses on the work you’d do when you […]
Even ’uncensored’ models can’t say what they want
Research / Even ’uncensored’ models can’t say what they want Research Workbench · April 2026 Cover image: Generated with Google Gemini · euphemismbench-flinch-profile.jpeg A safety-filtered pretrain can duck a charged word without refusing. It puts a fraction of the probability an open-data pretrain puts there. We call that gap the flinch, and we measured it […]
Monero Community Crowdfunding System
Donate to the Monero Project By donating to the following Monero address (General Fund), you are supporting the Monero Project. If you wish to donate to a specific proposal, please see Funding Required. 888tNkZrPN6JsEgekjMnABU4TBzc2Dt29EPAvkRxbANsAnjyPbb3iQ1YBRk1UXcdRsiKc9dhwMVgN5S9cQUiyoogDavup3H
Jujutsu megamerges for fun and profit
This article is written both for intermediate Jujutsu users and for Git users who are curious about Jujutsu. I’m a big Jujutsu user, and I’ve found myself relying more and more on what we in the JJ community colloquially call the “megamerge” workflow for my daily development. It’s surprisingly under-discussed outside of a handful of […]
OpenAI ad partner now selling ChatGPT ad placements based on “prompt relevance”
StackAdapt is quietly courting advertisers to test ads inside ChatGPT. The independent demand-side platform is dangling CPMs as low as $15 alongside discounted platform and management fees. The company is framing the push as early access to a new “discovery layer”—one that captures people in the middle of researching and comparing products on ChatGPT. According […]
Tim Cook to become Apple Executive Chairman. John Ternus to become CEO
April 20, 2026 PRESS RELEASE Tim Cook to become Apple Executive Chairman John Ternus to become Apple CEO CUPERTINO, CALIFORNIA Apple announced that Tim Cook will become executive chairman of Apple’s board of directors and John Ternus, senior vice president of Hardware Engineering, will become Apple’s next chief executive officer effective on September 1, 2026. The transition, which […]
AI Resistance Is Growing
As the internet chokes on ever more slop, the one thing that gives me hope is this: people seem to loathe AI, and are actively resisting it. This won’t be a long post, as I’m personally so tired of writing and thinking about AI at this point in time, but I do want to draw […]
F-35 is a masterpiece built for the wrong war
Think of a violin made by a master craftsman: beautiful, precise, capable of extraordinary performance, but impossible to produce quickly or cheaply. It takes time, rare expertise, and materials that cannot be sourced at scale. You would not equip an entire orchestra with instruments like that. Yet that is essentially what the United States has […]
Soul Player C64 – A real transformer running on a 1 MHz Commodore 64
A real transformer running on a 1 MHz Commodore 64. hellohey! nice to see you! i’m sadi hear you. i care about you. tell me a jokewhy did the bit flip? it was tired! Keep exchanges short – the model has a 20-token context window. See data/example_corpus.txt for a starter. python train.py data/example_corpus.txt This trains […]
We got 207 tok/s with Qwen3.5-27B on an RTX 3090
Open LLM inference, rewritten by hand for one specific chip at a time.Kernels, speculative decoding, and quantization, tailored per target.We don’t wait for better silicon. We rewrite the software. Inside the box Two projects today, more coming. Each one is a self-contained release with its own benchmarks and paper-style writeup. 01 · Megakernel Qwen3.5 […]