The three areas of need are: 1. Agent Authentication & Authorization 2. Prompt Injection-related attacks 3. Secure Agent Architecture For #1, Agent authentication can likely be put into existing stuff like okta etc. Authorization is the really hard part and will need to be

Some interesting data points about SWE agents

We are living in a timeline where a non-US company is keeping the original mission of OpenAI alive - truly open, frontier research that empowers all. It makes no sense. The most entertaining outcome is the most likely. DeepSeek-R1 not only open-sources a barrage of models but

Machines will train machines. Never bet against scaling. Never.

For friends of open source: imo the highest leverage thing you can do is help construct a high diversity of RL environments that help elicit LLM cognitive strategies. To build a gym of sorts. This is a highly parallelizable task, which favors a large community of collaborators.

Stanford presents: s1: Simple test-time scaling - Seeks the simplest approach to achieve test-time scaling and strong reasoning performance - Exceeds o1-preview on competition math questions by up to 27% (MATH and AIME24) - Model, data, and code are open-source

A funny thing about the deepseek-v3 training cost everyone is freaking out about is that they reported comparable training efficiency in the deepseek-v2 paper in May 2024. 172.8K hours * 14.8T tokens = 2.557M hours vs 2.788M hours

Excited to share that I’ve been working on scaling up diffusion language models at Inception. A new generation of LLMs with unprecedented capabilities is coming!

The best way to judge a forecaster is their track record. In 2021 Daniel Kokotajlo predicted o1-style models. I think we should all be very interested in the new predictions he's making in 2025! I've read it and highly recommend - it's thought provoking and stressfully plausible

WE'RE PREMIER LEAGUE CHAMPIONS! 🏆

I attended a vibe coding hackathon recently and used the chance to build a web app (with auth, payments, deploy, etc.). I tinker but I am not a web dev by background, so besides the app, I was very interested in what it's like to vibe code a full web app today. As such, I wrote

Wrapped up Stanford CS336 (Language Models from Scratch), taught with an amazing team Tatsunori Hashimoto Marcel Rød Neil Band Rohith Kuditipudi. Researchers are becoming detached from the technical details of how LMs work. In CS336, we try to fix that by having students build everything:

SF's flyering game is incredible right now. There's 1000s of flyers around the city that are completely unrelated and all completely sincere

Flow Matching (FM) is one of the hottest ideas in generative AI - and it’s everywhere at #ICML2025. But what is it? And why is it so elegant? 🤔 This thread is an animated, intuitive intro into (Variational) Flow Matching - no dense math required. Let's dive in! 🧵👇

Why you should stop working on RL research and instead work on product // The technology that unlocked the big scaling shift in AI is the internet, not transformers I think it's well known that data is the most important thing in AI, and also that researchers choose not to work

🆕 Everything you should know about Context Engineering youtube.com/watch?v=_IlTcW… Works like Chroma's Context Rot research and Drew Breunig's Context Fails show a lot of issues with naive long context usage: - Context Poisoning - Context Distraction - Context Confusion - Context

This is a big deal. It is the first large-scale demonstration of the advantage of real-time reinforcement learning. The recipe is scalable and requires no intervention in principle; the model can adapt forever as long as it is being used. There is no way to achieve similar

Interesting read: agentic browser makes same-origin policy irrelevant

The first research on the fundamentals of character training -- i.e. applying modern post training techniques to ingrain specific character traits into models. All models, datasets, code etc released. Really excited about this project! Sharan was a joy to work with.

It was great to help with this interactive tutorial on SAEs, what they can be used for, and how they work. Fantastic work by the team!