Radar / AI robotics / FineART
FineART: Scale AI and Hugging Face's open robotics dataset and plan-before-act VLA model
FineART is the largest open robot dataset with subtask labels: 40,543 episodes, 1,718 hours of footage, 151 tasks and 533,913 labeled steps. Alongside it, FineART-VLA is LeRobot's first open System 1-System 2 vision-language-action model, predicting subtasks in language before converting them into robot actions.
Why it matters
Subtask labels are the missing ingredient for robot policies that plan in language before acting, and the whole stack loads natively in Hugging Face's LeRobot ecosystem.
What you could build with it
A robotics startup building a warehouse picking arm could fine-tune FineART-VLA on its own subtask-annotated demos, giving the arm a planning step that narrates each pick before moving, which makes failures diagnosable instead of opaque.
Does it hold up?
Too early to judge: released 29 Oct with full LeRobot integration, but no independent robot deployments reported yet beyond the launch materials.
Built with FineART
- Scale AI and Hugging Face Release FineART Robotics Dataset and Modelarticle · Recaps the release: 40,543 episodes, subtask labels pairing timed action clips with language, and transfer across robot hardware.
- FineART-VLA model documentationhuggingface · LeRobot docs for FineART-VLA: how it differs from Pi05, the language-head training setup, and the install path.
- huggingface/lerobotgithub · LeRobot hosts the FineART-VLA policy, training recipes and the FineART dataset loader.
Learn more
First spotted on article: source.
More AI robotics
Bear Robotics KINO-2: demonstration-to-skill robot-learning platformBear Robotics launched KINO-2 on Oct 21 at GTC Berlin: a robot-learning platform that turns human task demonstrations…robotics · JEV 0.59Dream-1: robot foundation model that controls unseen bimanual robots zero-shot, at $3/hourDream Labs came out of stealth on 19 October with Dream-1, a robot foundation model offered through a public Dream API…robotics · JEV 0.52AGIBOT Innovation Day: GO-3 Preview, WITA-Omni, G2 Max, Link-U OS 1.0 and the Worken productivity metricShanghai embodied-AI firm AGIBOT unveiled its GO-3 Preview embodied foundation model, WITA-Omni interaction model…robotics · JEV 0.43
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.