Search: “papers”

If you maintain an AGENTS.md or a CLAUDE.md, this one is worth your time. (bookmark it) Researchers traced 94K development events across 557 agentic coding sessions, plus 690K file-level change record

This is one of the most effective ways to improve your agentic workflows. If you are building computer-use agents, this one is worth your time. Task Model Induction takes a raw recording of someone wo

Impressive research from Google on building better environments for agents. Training environments for agents are hand-built and go stale. The agent improves, the environment does not, and it's not abl

Does your context compactor actually save anything? If you built or tuned one, this one is worth your time. (bookmark it) Task completion is the standard way to validate compression, and it hides the

Interesting new research from IBM. If you pick models from benchmark deltas, some of that delta belongs to the phrasing rather than the model. BenchDrift generates meaning-preserving variations of ben

Four small architecture decisions can cost up to 47% of a model's long-context performance. New research from Ai2, Carnegie Mellon, and the University of Washington isolates them. Normalization, GQA,

New research from Meta. Agent harnesses are still mostly authored by hand. This makes it hard to tune robust agent harnesses for long-horizon tasks. In this new work, agents learn harness policies off

New research from Google DeepMind. (bookmark it) SkillSmith treats model weights as an additional modality the LLM reads natively. The augmented model ingests existing prefix weights alongside rich te

New work led by @FlemmingKondrup and @tomjiralerspong highlights an important vulnerability in chain-of-thought monitoring for agents, I highly recommend giving it a read. Link to the paper: https://a

// ReContext // Models now support 128K context windows and still fail to use evidence that is already in the prompt. Where is the gap? New paper introduces ReContext, a training-free inference harnes

further evidence that people shoveling money into AI are either bad at math or blind to risk Striking paper from Wharton. The big conclusion: AI must increase productivity 2.7x -- and quickly -- or te

Active Graph is the best, most 'correct' knowledge/context engine I've come across so far (and I've tried or at least researched most of them.) babyagi has ~200 citations, but 0 papers... i just publi

If you design production agent systems, this matters. Most devs accidentally let their framework defaults make critical architecture decisions without thinking it through. This paper shows you how to

// δ-mem: Efficient Online Memory for LLMs // One of the more elegant memory mechanisms I've seen this month. Most long-term memory work either inflates context or retrains the model. This paper shows

Wanna get a million views? Make stuff up. Take a tiny tiny bit of truth and distort it wildly. Consider the tweet below, 1.4M views. Take the chess thing. The paper that is linked doesn’t actually say

We’re excited to introduce KAME: Tandem Architecture for Enhancing Knowledge in Real-Time Speech-to-Speech Conversational AI, accepted at #ICASSP2026! 🐢 Blog https://pub.sakana.ai/kame/ Paper https://

Pay attention to this one, AI devs, especially if you're thinking about agentic commerce or any agent network where many agents share hosts. A correct route to a cold agent is still a failed request f

How do AI Agents spend your money? Most teams treat agent token costs as a rounding error even though the data says they shouldn't. New paper presents the first systematic study of how agents actually

Scaling massive monolithic LLMs continues to yield incredible results. But to truly unlock their ceiling, the next frontier is test-time compute and dynamic orchestration. Nature solves complex proble

Tool Attention Is All You Need // Tool Attention Is All You Need // New research proposes a practical fix for the hidden 'MCP tax.' The work introduces a dynamic tool gating mechanism built on an Inte

MIT researchers just replicated human muscles with AI-controlled fibers. Inside each fiber is a sealed tube of electrically charged liquid and a tiny electric pump. When the pump activates, one side c

ml-intern by @huggingface is wild 🔥 You drop a high-level prompt (“build the best scientific reasoning model” or “crush healthcare benchmarks”) and this open-source agent does the entire post-training

AI Insider: 'Adding a Human Makes Your Team Worse' Emad Mostaque | @EMostaque TIMESTAMPS : 00:00 The models too dangerous to release 10:06 Why physics needs axioms — and AI doesn't 11:43 The MIND fram

It’s time to go beyond language models. Introducing Odyssey-2 Max, our most powerful world model yet. It materially advances the SOTA in physical accuracy. This is a big step toward models that simula

Earlier this year Yann LeCun left Meta because Mark Zuckerberg wouldn't bet the company on JEPA. Last week his group dropped the first JEPA that actually trains end-to-end from raw pixels. 15 million

Intelligence per picojoule, with @itsclivetime and @dylan522p (0:00) Intro (1:22) What is codesign? (2:49) Codesign example: Swish vs ReLU (4:22) Are DeepSeek papers codesign? (6:45) Predicting where

A startup called Sabi just came out of stealth with a beanie that reads your thoughts. 70,000 to 100,000 miniature EEG sensors woven into the fabric. You put it on like a winter hat and type by imagin

Introducing DDTree: accelerates speculative decoding by drafting a tree with one block diffusion pass, then verifying multiple likely continuations together. Paper: https://liranringel.github.io/ddtre

Really interesting ideas are going to be increasingly at a premium as the cost of executing those ideas drops. (Our research and others shows AI is quite good at generating interesting ideas, but not

I am a Web3 Ambassador at World Liberty Financial. There are 12 of us on the team page. 4 are named Trump. 3 are named Witkoff. The page calls us 'the passionate minds shaping the future of finance.'

What if a model became the computer itself? NEW paper from Meta. (bookmark this one) What if the model wasn't just using the computer, but became the computer? New research from Meta AI and KAUST make

I just built my own wiki generator plugin for my agents. My agents can now generate wikis for anything I ask. One of my favorite wikis is called PaperWiki. This is a great example of what @karpathy de

If you’ve encountered garbled output like this while using GLM-5 or GLM-5.1 on our official service, the issue is now resolved. We've patched the underlying inference-side bugs and will be releasing a

Gemini 4 TikZ unicorn From GPT-4 to GPT-6. Sparks was a remarkably prescient paper that got a lot of pushback at the time, but absolutely sensed where the vibes were heading with LLMs based on a lot o

New impressive work from Microsoft and collaborators. And it's one of my favorite recent topics: RL environments. They show that Qwen3.6-35B-A3B goes from 41.1 to 53.3 on Terminal-Bench 2.1 by trainin

Introducing Multi-Agent Self-Supervision (MASS): Recursive self-improvement through collective intelligence Blog: https://pub.sakana.ai/mass/ As AI systems tackle increasingly open-ended research prob

More agents don't mean higher performance. There is a coordination bottleneck to consider. Not to mention the unnecessary costs. So how many of a multi-agent team's actions actually help it finish the

Super interesting work from Zoom and colleagues. If you maintain a hand-built coding harness, there are some great insights here. (bookmark it) They held the execution loop of a coding harness fixed a

If you build agent harnesses, this is important. Should you avoid subagents, or can they be useful? My thoughts as a harness builder: I remember using subagents in Claude Code, and I mostly found them

The state of public AI benchmarking is dire and is undermining our ability to understand how good AI is now. Most famous measures are maxed out, and, as this paper shows, the non-saturated benchmarks

Agents are coming for video production. @PooldayAI is going for this. You prompt it once, and it edits your real footage, picks the right generative model for each task, and checks its own work before

THE CURE IS THE PITCH Dario’s 'Pacing' Essay Is Quiet-Period Illegal Stock Promotion Wrapped in Regulatory Capture Dario Amodei's 'We Must Pace the Frontier' is not a safety paper. It is a pre-roadsho

If you run automated prompt or harness evolution, this one is worth your time. (bookmark it) Semantic-gradient prompt optimization improves an agent harness from execution feedback, but its local upda

So much discussion on the emergent behaviors of agent swarms. This is a great read from Google DeepMind if you are tracking this research topic. https://x.com/omarsar0/status/2095873020778991918?s=20

I highly encourage anyone interested in AI safety to read this. I remember discussing the importance of observability with my mates at @BlueDotImpact last year. This feels more important than ever URG

Impressive new work from Google DeepMind showing real-world applications of agents for scientific research. Impressive new paper from Google DeepMind. (bookmark it) It takes Co-Scientist out of simula