Yuanli Lingji's DM0.5 model has achieved state-of-the-art performance on the RoboDojo benchmark for embodied AI, demonstrating exceptional memory capabilities and a 19.34% average success rate. The model also excels in other evaluations like LIBERO and RoboTwin 2.0, showcasing strong generalization across diverse tasks including human imitation, fine manipulation, and stable operation in complex environments. Yuanli Lingji has open-sourced the DM0.5 model, its training framework, and downstream task workflows, enabling broader research and development in the field. AI
IMPACT Sets a new benchmark for embodied AI capabilities, particularly in memory and generalization, potentially accelerating enterprise adoption of advanced robotics.
RANK_REASON The cluster reports on a new state-of-the-art model release (DM0.5) for embodied AI, including benchmark results and open-sourcing details. [lever_c_demoted from frontier_release: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
