LeRobot v0.7.0: From Terabytes to Prompting Robots
Blog post from Hugging Face
LeRobot v0.7.0 expands the open-source robotics platform across data, training, inference, simulation, and hardware, supporting nearly 300 TB of Physical AI datasets hosted on the Hugging Face Hub. The release enables diskless dataset streaming from the Hub, storage buckets, and LanceDB; GPU-based image augmentation; and distributed training features such as FSDP2, HSDP, distributed checkpoints, mixed precision, EMA, and Trackio experiment tracking to accommodate multi-billion-parameter world action models. It introduces interactive language-driven robot control, allowing operators to change subtasks, ask visual questions, or assign autonomous goals during rollouts through text-capable policies or external vision-language planners. New supported models include world action models such as Cosmos 3, FLUX 3 Action, LaWAM, and Long-WAM, alongside vision-language-action models including G0.5, LingBot-VLA 2.0, DM0.5, and Wall-OSS-0.5. Remote inference lets robots use lightweight local computers while executing policies hosted on GPU servers, while a unified simulation framework improves evaluation and supports RoboLab benchmarks with confidence intervals. The update also adds support for YAM and reBot arms, enhanced Unitree G1 humanoid control using NVIDIA SONIC, typed codebase improvements, browser-based dataset access, and expanded documentation and community resources.
| Trend | Post Mentions | Total Month Mentions | Posts | Companies | MoM |
|---|---|---|---|---|---|
| Real-time | 6 | 2,990 | 636 | 190 | -28% |
| AI Model Fine-tuning | 5 | 613 | 130 | 56 | -29% |
Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.