🚀 NVIDIA continues to lead on open-sourcing pretraining data — Nemotron-CC-v2 has dropped! 👏 Congrats to Rabeeh Karimi Sanjeev Satheesh Pavlo Molchanov Kezhi Kong X. Dong Bryan Catanzaro Yejin Choi + many others! 🙏 A very loud thank you for citing our Physics of LMs, Part 3.1.

I'm excited to share that I'll be stepping into the role of Head of Cohere Labs. It's an honor and a responsibility to lead such an extraordinary group of researchers pushing the boundaries of AI research.

Meta Superintelligence Labs presents Language Self-Play For Data-Free Training

When we at OpenAI released o1-preview a year ago, it would think for seconds. Today, our best reasoning models can think for hours, browse the web, and write code. But there's a lot of room to push reasoning even further. I'm excited for what the next year will bring!

Just got the greenlight to share some work we did at Google DeepMind from over a year ago: We fine-tuned Gemini on thousands of the most toxic discussions on 4chan...and it just talked to us like a completely normal and nice language model. How? Our method, Generative Data

-2016 (classic era): focus on data efficiency 2017-2025 (pretraining era): focus on compute efficiency 2026-: focus on data efficiency (again) The standard Transformer paradigm is optimized for compute efficiency. As we look at data efficiency, we'll see very different design

We’ve received A LOT OF submissions this year 🤯🤯 and are excited to see so much interest! To ensure high-quality review, we are looking for more dedicated reviewers. If you'd like to help, please sign up here docs.google.com/forms/d/e/1FAI…

Spent an incredible two days in Zurich speaking with students at ETH Zurich. I shared a little bit of my journey as an entrepreneur and spent time inside the ETH AI Center, learning about what they've been working on. The questions from students reminded me just how much curiosity

The first ever Zurich Robotics event is tonight 18:00 ETH AI Center: Barnabas Gavin Cangan (Barnabas Gavin Cangan, ETHZ) on why robot hands are so hard and Caterina Caccavella (ZHAW / ETHZ) on bio-inspired active sensing. zurichai.ch/events/zurichr…

🇨🇳 China just dropped a new GPU: Fantasy III - Claims CUDA compatibility and Ray tracing support - 112GB+ of HBM memory for AI - positioning it as an all-purpose card for gaming, compute, and medical imaging. - For AI, the memory footprint is pitched as enough for 32B and 72B

Understanding the capabilities of AI models is important to me. To forecast how AI models might affect labor, we need methods to measure their real-world work abilities. That’s why we created GDPval.

Sharing our second Connectionism research post on Modular Manifolds, a mathematical approach to refining training at each layer of the neural network

📢 Early submission for ACL 2026 via ARR Oct cycle is available to support authors facing potential visa delays. 📝 Early invitation letters possible for submit-ready work (not acceptance guarantees). ⚠️ Preliminary work risks rejection & Jan-cycle ineligibility. #NLProc #ARR

Excited to announce the 9th FEVERworkshop collocated with #EACL2026 in Morocco: fever.ai/index.html with a new shared task on #factchecking image-text claims with evidence from the web using the AVerImaTeC dataset: arxiv.org/abs/2505.17978 Deadline 2nd of December! And...

🔍 Do vision-language models truly understand diagrams - or just leverage shortcuts? Excited to share our new paper: “Chimera: Diagnosing Shortcut Learning in Visual-Language Understanding” lead by Ziheng Chi & Yifan Hou Yifan Hou! 📄 Paper: arxiv.org/abs/2509.22437 ✨