Search: “papers”

🎨 Meet Qwen-Image-3.0 — the third generation of our foundational image generation model. If 1.0 was about 'Precision,' and 2.0 added 'Variety, Completeness, Beauty & Authenticity,' then 3.0 comes down

Highly recommended. I've often claimed there's huge alpha in building agent harnesses. Turns out harnesses are compositional generalizers. The RLM harness is an instance of this. This could lead to in

Today, we are introducing Inkling. Inkling reasons efficiently across text, image, and audio modalities. We are making the full weights available. https://thinkingmachines.ai/news/introducing-inkling/

Fable, turn my tweet into a thinkpiece (this was pretty funny): There has never been a better time to have opinions about artificial intelligence. I say this with some authority, because I am currentl

Everyone keeps asking me how to build a second brain or an LLM wiki. Here is the easiest setup I have found. I took my Wiki Builder skill, installed it into HyperAgent (@hyperagentapp) as a reusable s

Great writeup from the University of Oxford. It's a taxonomy of LLM-based agent limitations. Good read for anyone shipping with agents. Benchmark scores keep climbing, yet the same agent failures resu

Qwen publishes new work on RL coding agents. (bookmark it) The idea is to continually build a verification system that co-evolves with AI agents. LLMs suffer from all sorts of reward hacking issues. T

When does combining LLMs help? Great analysis on combining language models, measured across 67 models from 21 providers. Any policy that routes, votes, cascades, or runs a mixture of agents and then r

// The Consistency Illusion // Multi-agent debate can make agents agree on the final answer while their underlying reasoning stays misaligned. This work finds that consensus on the output hides disagr

I'll never forget @GaryMarcus' comment at last autumn's @MediaTechDem conference in MTL - something to the effect of 'what if Nvidia are just the people who sold shovels in the gold rush'? Striking pa

New MIT study. Code volume surges by 300%, but output increases by only 30%: The AI dividend meets an awkward reality Autonomous AI coding agents raised commits by 180%, but releases rose only 30%. Th

Nice primer on post-training reasoning data. (bookmark it) This is one of the first primers to pull the scattered post-training reasoning-data literature into one place, synthesizing over 150 public s

Highlighting the new WebGPU backend in llama.cpp/ggml The work to bring full-fledged WebGPU support in llama.cpp started about an year and a half ago. It has been lead by @reeselevine and team at USCS

Can we transform the Hugging Face Hub🤗—with its enormous sea of artifacts—into a self-evolving discovery machine? WE CAN. Introducing ArtifactLinker! 🚀🔥 We are fundamentally rethinking how we interact

The pure LLM debate - which I had for many years, here and elsewhere - is indeed no longer relevant. Why? Because I won; nobody uses pure LLMs anymore. Nowadays all deployed objects are neurosymbolic,

Big move from arXiv: a one-year ban for authors who submit AI-generated content without proper checking. This is not about banning AI from academia. AI can be extremely useful. It can help us write be

Pay attention to this one if you build research or knowledge-work agents. Most research-agent systems produce uniform outputs regardless of who is driving them. This new work, NanoResearch, argues tha

Am old enough to remember when @GeoffreyHinton told me I was stupid for saying that LLMs regurgitate training data. He was wrong. LLM regurgitation is now one of the best-established findings in the f

// LLMs Improving LLMs // Interesting progress the past of couple of weeks around self-improving AI agents. If autoresearch was interesting, you will like this read. (bookmark it) We've been hand-tuni

This seems like a critical reason to open up about AI use in academia. Scholars are using old AI models, badly, and not talking about it. New models hallucinate very few citations, and good agentic ha

The human brain🧠 is incredibly efficient because it only activates the specific neurons needed for a thought. Modern LLMs naturally try to do this too (> 95% of neurons in feedforward layers stay sile

// HeavySkill // One of the cleaner takes on agentic harness design I've read. They argue that what actually drives agent harness performance is not the orchestration code. It's a single inner skill:

Are AIs about to subjugate humanity? In the debate about catastrophic AI risk, evolutionary scenarios receive far too little attention compared to largely speculative arguments about “instrumental con

Claude Opus 4.7 just implemented an AlphaZero-style self-play pipeline from scratch. It did this on consumer hardware in three hours, then beat the Pascal Pons solver 7 of 8 as first-mover on Connect

'AI should elevate your thinking, not replace it.' I don't disagree, but the issue is that current LLMs are not really trained to support that out of the box. I've solved this by building my own agent

For the past few years, humans have been doing “prompt engineering” to coax the best performance out of different LLMs. In this work, we explored what happens if we train an AI to do that job instead.

Looking forward to the work coming out of @IneffableLabs 🇬🇧 Largest EU/UK raise ever 😮 Beat the previous UK largest seed raise (101m Stability AI!) by 999m 🚀 We have seen amazing things on our self-le

Very cool analysis of the submissions to a major management journal that shows how much the system of science, built for humans, is under strain as a result of AI. AI can be used to do better science

Here is a very common problem when building complex agents. Long-horizon agents (in particular) fail in two ways: the decision-maker can't decompose well, or the skill library goes stale. This new res

👀 The most interesting people in the most exotic places can occasionally be persuaded to talk about NFTs, digital collectibles, and even @CandyDigital! Try it sometime just in case it catches on… 🔥🚀📈

Alpha Eval: Agents Making Evals as a Multi-Player Game This diagram and blurb is largely a research riff with Claude on building data generation systems to get closer to the holy grail of self-improvi

Bet this happens with Navier Stokes and it’s going to be something not even related to PDEs that solves it 23 years old with no advanced mathematics training solves Erdős problem with ChatGPT Pro. 'Wh

5B tokens in ml-intern in 48h 😅😅😅 we burned through 5 billion tokens in 48h. turns out giving everyone unlimited access to the most expensive model on the planet is not a great business strategy 🙈 So

If you can’t stop small teams from using your API for distillation, then you’re definitely not stopping criminals, biohackers, or adversarial states from using AI through it. Those actors are much mor

A couple of months ago, I announced that I was partway through implementing a simple, readable AlphaFold2 in pure PyTorch, inspired by @karpathy's minGPT. Today, I'm happy to share minAlphaFold2 - the

🚨 Are we witnessing the automation of AI research? @HuggingFace just unveiled 'ML-Intern' and my mind is BLOWN 🤯 It’s an open-source pipeline that replicates the exact daily loop of an ML researcher.

🔥 Emad Mostaque on Doom Debates 🔥 @EMostaque helped kick off the modern AI revolution as the head of Stability AI, the company behind Stable Diffusion. But unlike most AI CEOs, he doesn't sugarcoat th

I kinda think it would be interesting to have a structure where founders take 2 and 20 2% management fee on capital raised for running the company 20% of any exit Single magic share with lots of votin

Let me say this clearly: LLMs cannot feel emotions. Emotions are evolutionary mechanisms. They push us to avoid danger or approach what is beneficial. We experience emotions because we are alive, and

Source: https://news.mit.edu/2026/new-type-electrically-driven-artificial-muscle-fiber-0409?utm_source=robotnews.therundown.ai&utm_medium=newsletter&utm_campaign=unitree-s-cheapest-humanoid-goes-globa

This thread discusses how advances in AI tooling and accessible models have lowered barriers to entry for startups, enabling small teams to leverage capabilities previously requiring specialized ML ex

I tested @huggingface ml-intern, given the prompt 'Fine-tune a Segment Anything Model (SAM) on a useful medical dataset. Train the model, and provide a comprehensive tutorial in a Jupyter Notebook fil

We are happy to share early results from Logos, our novel first-principles augmented intelligence system, that has enabled insightful results across domains. We start with series of results in physics

Localmaxxing.com is a platform in private testing phase with plans for public release the same week this announcement was made, designed to simplify benchmark sharing by eliminating the need for users

I am not convinced that we should be comfortable calling 'problem solving' or 'judgement' or whatever as skills that are impossible for AI to do well. Like any other skill, there are humans who are re

We only have spotty information about this very important topic. It suggests AI can be good at diagnosis, but the real world doesn't always match the experiments. https://x.com/emollick/status/1980474