On August 31, Interface News learned that Zhiyuan officially released the AGIBOT WORLD 2026 3rd Edition Embodied Intelligent Reinforcement Learning Open Source Data Set. The first batch opened 11,430 real interaction trajectories, covering real scenarios such as industry and home, including 14 types of interactive tasks such as plugging in a network cable and opening a key. The data includes over 98,000 fine-grained key state labels, supports training reward models with different dimensions, and provides a reusable empirical data foundation for real machine reinforcement learning.
Zhitongcaijing · 2d ago
Share
Listen to the news
On August 31, Interface News learned that Zhiyuan officially released the AGIBOT WORLD 2026 3rd Edition Embodied Intelligent Reinforcement Learning Open Source Data Set. The first batch opened 11,430 real interaction trajectories, covering real scenarios such as industry and home, including 14 types of interactive tasks such as plugging in a network cable and opening a key. The data includes over 98,000 fine-grained key state labels, supports training reward models with different dimensions, and provides a reusable empirical data foundation for real machine reinforcement learning.
Disclaimer:Webull uses external vendor Google Translation Service for news translations where we endeavour to ensure these are correct, however, we recommend that you please double-check this information accordingly. Webull is not responsible for translation errors or issues.
What's Trending