AK

for daily papers, authors can directly submit them here: huggingface.co/papers/submit

Reachy Mini was everywhere at GTC 😍 Why? Because it’s the perfect interface to interact with your local, private agents (hey, Nemoclaw!) Thanks to our partners @nvidia , @Dell, @ASUS , @Lenovo, @HP and @msitweets for showing the power of private agents running locally on GB10. So much fun to be able to talk to them, showing them what you want instead of reaching for an app and typing, uploading images. 🦞+ 🤖 = Reachy Pinchy?

V-JEPA 2.1 is out Unlocking Dense Features in Video Self-Supervised Learning paper: huggingface.co/papers/2603.14…

We just submitted APEX-Agents, APEX-1 and ACE to @evaluatingevals on @huggingface, an OSS initiative to standardize evals and try to reduce the noise in benchmarking.

Immediately makes HOI-Blender result much more photorealistic lol

Alignment Makes Language Models Normative, Not Descriptive We compared 120 base–aligned model pairs on 10,000+ human decisions. Base models outperform aligned models nearly 10:1 in strategic games. Alignment learns what humans should do; base models learn what they actually do.

🧩 We just saved $100,000,000 in 4 days making this AI movie Introducing Higgsfield Original Series - world's first complete AI streaming platform showcasing next generation AI filmmakers. Discover AI films and series and vote on which of the teasers gets continued. Revolution in AI filmmaking just happened - Ep. 1 of Arena Zero debuted on Higgsfield Original Series OST: Dirty Ctrl - DEM

Efficient Exploration at Scale paper: huggingface.co/papers/2603.17…

Nvidia just crossed Google as the biggest org on @huggingface with 3,881 team members on the hub. I'm officially calling it: Nvidia is the new American king of open-source AI!

Our biggest open-source repos are getting overwhelmed by AI slop which literally makes Github unusable (~a new pull request every 3 minutes). Fun new challenges in an agentic world!

MetaClaw Just Talk An Agent That Meta-Learns and Evolves in the Wild paper: huggingface.co/papers/2603.17…

MetaClaw: Just Talk Your personal agent evolves automatically from every conversation. MetaClaw meta-learns in the wild using cloud LoRA and skill synthesis, scheduling weight updates during sleep or idle time with zero downtime.

Video-CoE Reinforcing Video Event Prediction via Chain of Events paper: huggingface.co/papers/2603.14…

Thanks @_akhaliq for sharing our work! If you are interested in V-Co, please see our thread for more details: x.com/hanlin_hl/stat…

lol

MiroThinker-1.7 & H1 is now the #1 paper on Hugging Face Daily! 🏆 A huge thank you @_akhaliq to the Hugging Face community @huggingface for the support. Our latest reasoning agents are designed for reliability through verifiable intelligence. Open weights and model repositories are live. Let's build the future of reasoning together! huggingface.co/papers/date/20… #MiroMind #OpenSource #DeepResearch

Video-CoE Alibaba researchers introduce a Chain of Events paradigm that breaks videos into temporal sequences to enhance logical reasoning for future event prediction, establishing a new state-of-the-art on video event prediction.

NEW SOTA OCR MODEL DROPPED Congrats to @VikParuchuri and team for releasing Chandra OCR 2! - 85.9% on olmocr bench, making it first place 🏆 - 90+ language support - 4B model - Full layout information - Extracts + captions images and diagrams - Strong handwriting, math, form, table support Compare every OCR model on the hub and choose the one adapted to your needs 👇

love it

Kinema4D Introducing an action-conditioned 4D generative robotic simulator that disentangles precise kinematic control from environmental dynamics, enabling physically-plausible simulation of complex robot-world interactions.