> We use agentic misalignment as a case study to highlight some of the… more
49 links
> Just a few months ago, AI-generated security bug reports to open source… more
> We introduce Natural Language Autoencoders (NLAs), an unsupervised method for… more
> ds4.c is a small native inference engine for DeepSeek V4 Flash. It is… more
> [...] We evaluate AgentFlow on TerminalBench-2 with Claude Opus 4.6 and on… more
> As we wrote in the Project Glasswing announcement, we do not plan to make… more
> Mythos Preview has already found thousands of high-severity vulnerabilities,… more
> Abstract:Can a large language model (LLM) improve at code generation using… more
> Today, we’re announcing PrismML, an AI lab building the most concentrated… more
> A from-scratch PyTorch implementation of TurboQuant (ICLR 2026), Google's… more
> Investigating this, we found that: more
> We've achieved 10x data efficiency with NanoGPT Slowrun within a few weeks.… more
> We describe the persona selection model (PSM): the idea that LLMs learn to… more
> AI GameStore is a scalable open-ended AI evaluation platform that transforms… more
> This is a brief guide to my new art project microgpt, a single file of 200… more
> The Claude C Compiler is a milestone, showing progress at a different level.… more
> With its rollout, GPT-4o showed it was not just for generating dinner recipes… more
> As you can see, there is no real consensus on the “best solution” to the… more
> In July 2025, the Justice Department announced it would not make any… more
> Last week I hinted at a demo I had seen from a team implementing what Dan… more
> tl;dr Argumate on Tumblr found you can sometimes access the base model behind… more
> A short introduction to RLHF and post-training focused on language models. more
> LLMs are here to stay in our social spaces - there are already 20+ agents on… more
> LLMs are kind of like sails in that left free flowing they're completely… more
> Kaggle is excited to announce the release of Werewolf in the Game Arena, our… more
> This approach has become standard practice — Claude Code now automatically… more
> Pi is written by Mario Zechner and unlike Peter, who aims for “sci-fi with a… more
Someone made a demo that lets you play Zork, except you can talk to it using normal English, like an LLM. I like to lead it around by asking questions: "What's in the mailbox?" "What's behind the house?"
> [L]ike many others I rapidly went from about 80% manual+autocomplete coding… more
> Introducing co-do.xyz [Source] - a demo and an experiment (with no… more
> I like to think of language models like ChatGPT as a calculator for words.
> This is reflected in their name: a “language model” implies that they are tools for working with language. [...]
> Want them to work with specific facts? Paste those [...] as part of your original prompt!
> [...] today's frontier LLM research is not about building animals. It is… more
> If you can substitute "hungry ghost trapped in a jar" for "AI" in a sentence… more
> To say that Clawdbot has fundamentally altered my perspective of what it… more
> Recently I ran an experiment where I built agents on top of Opus 4.5 and… more
> In our new "AI/ML Product Management" class, the "pre-case" submissions… more
> After recovering, Tan joined online support groups for other survivors of AI… more
> Last month, the Washington-based company launched a satellite with an Nvidia… more
> So what's an old guy yelling at Claudes going to do? He's going to write his… more
> [L]awyers say that LLMs are a long way from reasoning well enough to replace… more
> I recently came across JustHTML, a new Python library for parsing HTML… more
> LLMs are useful because they generalize so well. But can you have too much of… more
> In mid-September 2025, we detected suspicious activity that later… more
> A recent study from Harvard University caught my attention and provides… more
> Have you ever asked an AI model what’s on its mind? Or to explain how it came… more
This is a crazy fever-dream of a post about alpha software and I recommend… more
> Talking to CaBot was both empowering and unnerving. I felt as though I could… more