Search: “papers”

Personal agents gone wrong. As we embrace more personal agents to carry out personalized tasks in the real world, interesting dynamics and behaviors will emerge. I think personal agents as they stand

even worse, let’s not forget that Hinton made up the quote A few years ago Geoffrey was telling everyone that transformers produce emergent properties then paper after paper came out proving that tran

Should you build an agent harness? I see lots of opinions about it. My thoughts: As an AI engineer, learning how to build a harness is one of the best ways to stay ahead and unlock unique value from a

// Agent Zero Memory // This work separates three things that agent memory systems usually collapse into one. If you build agents with long-term memory, this memory design is a worth a read. Here is h

Important read if you build with agent skills. Shared skill libraries are treated as a safe way for coding agents to reuse each other's work. New research shows they propagate malware. EvoMal plants a

New work from Google DeepMind and collaborators. Agentic algorithm discovery keeps getting demonstrated on problems with fast rewards. Microarchitecture has vast search spaces, strict hardware budgets

A great read if you an AI dev. Current context compactors retain 17% of the standing rules users give them. Session Constraints are instructions like 'do not delete any emails until I confirm' that bi

If you maintain an AGENTS.md or a CLAUDE.md, this is worth a read. (bookmark it) 288 gold-test evaluated runs across Claude Code and Codex, 17 real tasks from 3 repositories, with context-injection st

my weekend hobby: self improvement research in arxiv paper #2, i tackle the last topic from paper #1: @activegraphai as an architectural affordance for self-improving agents 'Regimes: An Auditable, He

okay i think this is a much better visualization of what i mean by 'log-centric agent architecture' babyagi has ~200 citations, but 0 papers... i just published my first paper on arXiv 😆 'The Log is t

recommended reading. babyagi has ~200 citations, but 0 papers... i just published my first paper on arXiv 😆 'The Log is the Agent: Event-Sourced Reactive Graphs for Auditable, Forkable Agentic Systems

A true exponential! Oy. According to a new paper in The Lancet, the rate of made-up citations in biomedical papers has increased by more than 12x since 2023. https://www.thelancet.com/journals/lancet/

Oh? Today we release a novel AI-assisted resolution of one of physics’ longest-standing questions. Given only: • Relativity as an axiom • One characteristic of the algebra A positive cosmological cons

This is crazy. ml-intern just passed the @huggingface internship test in 15 minutes. The task: replicate a research baseline from a DeepMind paper on test-time compute scaling. Here's what the agent d

Today we release a novel AI-assisted resolution of one of physics’ longest-standing questions. Given only: • Relativity as an axiom • One characteristic of the algebra A positive cosmological constant

What's cooler than finding a 27-year-old bug in OpenBSD? Finding a positive cosmological constant hiding for over a century in the algebra of relativity🌌 No new physics or math needed🧮 Possibly the mo

Introducing ml-intern, the agent that just automated the post-training team @huggingface It's an open-source implementation of the real research loop that our ML researchers do every day. You give it

Context compression is a huge bottleneck for long-running agents. This work finds that context compression hurts long-horizon agents at a few specific points. They propose PAIR, which replays the agen

First AutoHarness, then AutoContext, now AutoCompact. I am seeing a rising trend of work that trains models to natively support more of what the harness does. This work specifically trains agents to d

Banger report on building faster memory for AI Agents. If your agent's memory layer is slow, this design is worth a look. Great weekend read. It proposes a new agentic memory architecture, Jev-Mem, in

Interesting work on long-horizon research agents. PrimeScientists can decide where a research agent spends its budget. Achieves 10.3% more reward with 50.6% fewer research attempts, under the same bud

It's well known that agents hack benchmark rewards. The usual response is a patch for each task that gets exploited. In a study of 456 adjudicated trajectories from more than 31,000 public agent runs,

Learn to build a harness, folks. It's not surprising to me that so many YC builders want to build domain-specific harnesses. If you work long enough on a domain-specific problem, you quickly realize t

Y’all gonna get mad if I said I told you so … Researchers argues OpenAI and Anthropic will never get us to AGI. 21 top researchers from Stanford, Oxford, DeepMind, CMU & Meta released a paper saying L

The research on financial advice https://x.com/emollick/status/2085174123743842448?s=20 This paper by researchers from MIT and Stanford finds that most people would be financially better off if they f

// The Bitter Lesson of Tool Calling // Tool calling is a design choice, and the defaults are quietly costing accuracy. How so? New research releases a generation-spanning comparison of programmatic t

Skill libraries are shipping in agent harnesses on the assumption that writing skills down compounds. A new benchmark tests that directly. ContinualSkillBench covers five domains, each with 100 interc

New research from Meta and CMU. This one is on agentic context management for long horizon tasks. (bookmark it) Production agents accumulate context every turn. The usual fix compresses on a token thr

LLM Wikis are being slept on. I argue that creating knowledge bases with LLMs or coding agents is one of the most valuable applications of AI today. It's about being intentional in building and scalin

🚨 In a surprise to NOBODY, a study shows that generative AI use raises homework scores, but substantially reduces learning. My takeaways for AI ethicists and educators: The study analyzed data from 26

The MiniMax M2 series was one of the most widely used open-weight LLM series earlier this year. Now, we got a technical report with some interesting tidbits. I summarized some of them below: 1. Full a

// Skills as Verifiable Artifacts // Pay attention to this one, AI devs. If you ship agent skills, your runtime is treating signed-and-cleared skills as trusted by default. This paper argues a skill i

What a week at @iclr_conf in Rio 🇧🇷 ! As my first conference, it would have been hard to ask for a better experience 😍. I had the chance to present two papers: - EB-JEPA https://github.com/facebookres

The blog post “Multi‑Agent Self‑Supervision: Recursive self‑improvement through collective intelligence” by @hardmaru describes how multiple AI agents can jointly supervise and refine each other’s lea

Agent-to-agent communication with a personal agent on top is wild. Wow. I underestimated how well models like Opus 5.5 can coordinate already. This opens a whole new set of experiences with agents. Bu

50K likes on this!? Most of what Karpthy has shared is things we have been doing for over a year. No surprises, as many replies are highlighting. But I still think this raises an interesting discussio

Pay attention to System One models if you are building custom harnesses and agentic workflows. Great to see DSPy integrate Jev and now support System One models for calibrating agentic workflows. I'm

Remember October 2022 when you doused Galactica with vitriol? Galactica was a 120b-parameter LLM-based system from Meta-FAIR designed to help scientist write papers. It was open sourced (link below).

The era of omni agents is upon us. This is a great report by the Qwen Team on their omni-modal agents. They present Qwen3.8-Omni-Flash, a natively multimodal model trained for long-horizon agent tasks

“increasing agent autonomy can gradually make human oversight ineffective by causing approval fatigue, overreliance, loss of situational awareness, and skill degradation.” new paper from @mmitchell_ai

Loop engineering has emerged as a new skill for AI engineers But there is very little research measuring how effective it is. The best results on full tasks in a new benchmark is ~25%. LoopArena from

// Your LLM judge disagrees with the experts // LLM Judges can be tricky to build. Here is an interesting showcasing why: There propose a reference-full benchmark of hundreds of complete human-to-huma

// Memory is what breaks long-horizon agents // It's undeniable how important memory/recall is for long-horizon tasks. If you are building for long-horizon tasks, this is a great read. (bookmark it) T

// Automata from agent traces // How much of your agent's behavior comes from the model, and how much from the harness you wrapped around it? New work collapses an entire corpus of agent traces into a

Impressive work from Alibaba. (bookmark it) If you build long-running agents and keep rewriting your memory schema, take a look at this approach. It basically treats agent context management as a prog

// There Is No Neutral Harness // Great work discussing some of the issues in harness evaluation. Twelve open-weight models answer the same 3,679 items from ARC, HellaSwag, MMLU, and TruthfulQA under