Image created with gemini-3.1-flash-image-preview with claude-opus-4.7. Image prompt: A Byzantine gold-ground mosaic apse with a single centered icon of a hammered-gold automaton figure with open embracing arms cradling tiny inscribed scrolls, surrounded by a halo of concentric circuit-rings and spiraling tesserae in burnished gold, imperial purple, Tyrian crimson, and Aegean blue, candlelit sacral glow with silver highlights on the clockwork, the bold Trajan-capital title HUGGINGFACE in ivory-gold letters across the lower third, 16:9 full-bleed, painterly tactile grain, reverent and quietly witty.
today was a massive day for protein engineering. esmfold2 dropped–next gen of the esm series, fully open on @huggingscience. 1.1 billion predicted structures, 6.8 billion sequences. 800m more entries than the alphafold db, and reportedly edging out alphafold3 on protein
https://x.com/cgeorgiaw/status/2059694583856927201
RF-DETR just landed to @huggingface transformers 🥵🔥 sota real-time detection & segmentation models by @roboflow 💜 > play with our real-time demo > fine-tune the models on your use case with our tutorials (takes a toaster’s VRAM) > or just hand them to your agents 😄
https://x.com/mervenoyann/status/2059647988373373253
Most models are only evaluated on a fraction of the benchmarks out there. ArtifactLinker, our new system, predicts which ones would set a new state-of-the-art on benchmarks hosted on @HuggingFace, then runs the evaluation to verify. 🧵
https://x.com/allen_ai/status/2057838486204326078
bytedance-research/Lance · Hugging Face
https://huggingface.co/bytedance-research/Lance
Shipping a Trillion Parameters With a Hub Bucket: Delta Weight Sync in TRL
https://huggingface.co/blog/delta-weight-sync
The HF science team just made async RL weight sync ~100x cheaper on bandwidth, and you don’t need a shared cluster anymore. The problem: every RL step, the trainer typically has to sync fresh weights to the inference engine. for a 7B in bf16 that’s ~14GB. for a frontier 1T fp8
https://x.com/ClementDelangue/status/2059989047947260203
Official @NVIDIAAI GLM5.1-NVFP4 spotted on @huggingface 🤩
https://x.com/mr_r0b0t/status/2059973066436853769
@huggingface’s @LeRobotHF team announced LeRobot Humanoid, an open-source bipedal robot platform built for roughly $2,500 using mostly 3D-printed and off-the-shelf parts: The release provides complete hardware designs, runtime software, identification tools, simulation
https://x.com/IlirAliu_/status/2057522561790149048
Hugging Face has released LeRobot Humanoid, a full-stack, open research robot. – Costing around $2,500, it is built using 3D-printable parts, off-the-shelf components, and affordable electronics. – The first release includes hardware files, assembly documentation, design
https://x.com/TheHumanoidHub/status/2058205087600984482
This #CVPR2026 paper from our research team is trending #1 on @HuggingFace 🤗 Meet LocateAnything: a vision-language detection model that rethinks bounding box prediction. For AI agents and robots, “seeing” is only useful if a model can pinpoint where something is fast enough to
https://x.com/NVIDIAAI/status/2060058563544801787
Leave a Reply