Search: “papers”

Small models are cheap to run, but expensive to adapt. The hard part is not only fine-tuning. It is the surrounding loop that involves collecting data, diagnosing failures, building evals, avoiding re

SONIC training code + Finetuning checkpoint + VLA data collection scripts are open-sourced. Little easter egg on the GEAR-SONIC website too :) https://github.com/NVlabs/GR00T-WholeBodyControl SONIC is

Clem Delangue (or someone in their network) shared excitement about successfully running GLM 5.1 locally, noting it appears to be functional and effective. The post highlights the cost advantage of ru

New post: We show that small, cheap models can detect the flagship Mythos FreeBSD zero-day (CVE-2026-4747) using a simple harness we call nano-analyzer Models down to 3.6B active params (including ope

Cursor AI shared observations on X validating that multi-agent architectures demonstrate superior performance when tackling novel problems that fall outside the distribution of training data. The post

Hermes Agent Tip💡 Hermes supports dedicated auxiliary models for eight task types: 1. vision 2. web_extract 3. compression 4. session_search 5. approval 6. skills_hub 7. mcp 8. flush_memories Each tas

I like this view from the brilliant Terence Tao. I have always believed that AI could help to augment us. We built and train agents like that today. But all forms of intelligence can and should be pos

Chandra-OCR-2 is an optical character recognition model developed by DataLab, available on Hugging Face at datalab-to/chandra-ocr-2. The model was highlighted by Hugging Face CEO Clément Delangue, wit

It is notable that we are all debating exactly which markdown files are most important to feed AI (skills, memory, tool instructions) and in which order to feed them to get the best output. Feels that

(thinking traces are always summaries of the internal reasoning, since the actual internal chain of thought is a trade secret, but that doesn't mean they aren't useful when summarized, especially when

Neat experiment finds AI fact checks are rated as more helpful & less ideological than human ones 'LLM-generated Community Notes can achieve broader cross-ideological acceptance than human-written not

time to help builders own their intelligence, open everything 🤝 the data produced and Experiential Memory gained from every agent interaction is the one of the most valuable things you can own to impr

Our Lab just posted a new research report from Zimran Ahmed about how the game industry is adapting to AI. He spoke to people at 20 different studios and found a wide range of approaches to adapt (or

Things that make the jagged intelligence of AI harder to deal with than the jaggedness of humans: 1) Weaknesses are not always intuitive or identifiable in advanced 2) All LLMs have similar weaknesses

AI is jagged, but I think sometimes it is easy to overly focus on that. The generalness is a surprise too! LLMs may be optimized for verifiable fields like coding, but AI is also not bad at corporate

I was unable to retrieve the specific content of the referenced X (Twitter) post (status ID 2042355048311575010), as it is not accessible through web search results and X/Twitter requires a login t...

Hallucinations remain in LLMs, but note that over centuries we have developed complicated, successful machines that take uncertain output from unreliable sources & reduce the risk of errors. We call t

Sebastian Raschka (rasbt) used Randal Olson's Tufte Test tool, developed by Goodeye Labs, to improve the visual quality of a machine learning benchmark plot. The Tufte Test encodes seven of Tufte'...

🧬 🤗 A new atlas of microproteins in the human brain is now @huggingface! It brings together more than 1,000 microproteins, including one that’s reduced in Alzheimer’s and appears to play a role in how

Introducing QwenGyre, bringing elastic RL to extreme-long-horizon (XLong) agent training. ✨ It launches with two core designs: 🥳 - Elastic Scheduler: reallocates GPUs between rollout & training as dem

Introducing RSI-Anything: Unlocking Recursive Self-Improvement with World Models 🚀 Can AI learn to improve itself by predicting the consequences of its actions? Built on JEPA-Anything, RSI-Anything us

AI solved open mathematical problems. But the biggest surprise came afterwards. A team at Google DeepMind has used AI to resolve 9 open Erdős problems, including questions that had remained unsolved f

“The field of artificial life is a very long-term bet on intelligence.” 🐙 Our former intern Akarsh Kumar recently joined @MLStreetTalk to discuss the intersection of evolution and automated search for

TRL v1.15 is out, and it’s an absolute banger of a release for memory-efficient post-training. The main change: SFT, DPO, KTO, GRPO, RLOO and Distillation now use a fused LM head by default. Instead o

The largest agentic LLM Inference dataset has arrived! 206B tokens · 12,002 sessions · 1,186,582 LLM requests · 1,213,347 tool calls The traces include tool names, arguments, durations, and more. Expl

Language is not fundamental to human consciousness. In terms of evolution, it arrived only recently on our timeline. Consciousness evolved over millions of years through sensory awareness, emotion, an

Even if AIs become conscious, that doesn't imply that humanity should hand over control to them. https://eigenism.org NEW: According to a bombshell report in the New York Times, Anthropic co-founder C

Kudos to Pope Leo. I am disturbed by how some of my fellow AI researchers keep trying to elevate the moral standing of checkpoints and harnesses. I hope the rest of us on Team Meatbag reject this move

Today we're unveiling Trillium Labs @trillium_labs, a new non-profit to foster the open science of frontier AI. We're building open post-training recipes and will expand into open infra to study RSI,

Agents write application code, but you still get paged at 2am when it breaks. We wanted to know whether AI could handle that part of the job too. Introducing Incident Arena: a benchmark that puts codi

Introducing 'Scaling In-Context Imitation Learning' (SAIL) to be presented at #IROS2026. This work is a collaboration between Sakana AI and the University of Tokyo. Blog: https://pub.sakana.ai/sail Wh

Automated AI theorem proving has moved the frontier: The holy grail is not an AI that can produce an endless pile of true theorems. It’s an AI that can discover mathematics, and build each discovery i

blank cells in tables can mess up agent decision making many document OCR tools struggle with reading multiple blank cells in a document and cause values to be shifted from the original column and row

Today we establish that AI is as good as expert human tutors for immediate GRE learning gains (p=.015, n=2,383 students, study conducted July-Sep 2026). One hour with an expert GRE tutor: $75. One equ

glance-vlm speedlab is now open source! read: https://glance.yohei.me/speed/ try: https://github.com/yoheinakajima/glance-speedlab turn any webcam into multiple live AI detectors (emotion, count, obje

There’s an enormous amount of open training data on Hugging Face. What does it take to make it work together 🤗? For Marin’s 535B run, we built on 25T tokens from 152 datasets with licenses permitting

Recommended if you work on agent memory. Retrieval is the hardest memory problem in most harnesses because agents keep pulling stale context. AML tests this with a coding track of 150 software tasks,

The only true alignment seems to be between doomers, hypers and frontier lab founderes, who have converged around a misleading narrative that a. completely glosses over the important technical advance

I said 'auto-regressive LLMs, in and of themselves, will not lead human-level AI' That statement is still totally true. First, the reasoning abilities of current AI systems are based non-auto-regressi

Live now: our Inference Engineering Track from AI Engineer World's Fair 2026. A benchmark tool told to run 200 queries a second that ran 38 and reported 200. A model that answered one prompt in a thou

Today’s LLMs still write like typewriters: one token at a time. This sequential process creates a hard inference bottleneck. We're introducing Uno, a diffusion-augmented LLM that delivers autoregressi

Nice work from Adobe. It's standard practice to store a fixed reference answer for every eval case. When the underlying data changes daily, that stored answer goes stale. Adobe researchers write each

On building an agent harness from scratch. Got so many questions about where to get started. My short guide (feed it to your agent): If you really want to learn harnesses well, it's worth building one

Day 3 dilemma Another 3-5,000 word blog on something interesting or should I strip it back to 800-1,000 words for wider readership? Should I discuss the equations of AI and get a bit technical or keep

Slowdown or no, OpenAI and Anthropic are like someone trying to sell billions of bottles of Perrier at £1,000 each. OpenAI and Anthropic are trying to sell us Perrier water at £1,000 a bottle in a wor

Do large language models actually understand the world, or are they just very good at pretending? Does it even matter? Our recent special issue in the Royal Society, “World Models in Natural and Artif