Research
Hugging Face Papers details AgenticRoboRL, a post-training method for robot planners
The framework post-trains a 9B planner with supervised fine-tuning and multi-turn RL, raising LIBERO-PRO success from 18.1% to 71.0% without external memory. The authors say they will release the framework and planner checkpoints.
Every outlet that ran it 1 article
Hugging Face PapersResearchLead
AgenticRoboRL: Bringing Coding-Agent Post-Training to Agentic Robotics