VLA Memory Goes Native, a World Model Runs 720p at 60fps

Featured

• Scaling Mixture-of-Experts Video Pretraining for Embodied Intelligence score 10
  入选 HF Daily Papers; HF 热度: 42 upvotes (+4); 有代码实现; 关键词(5): scaling, pretraining, MoE, embodied, open-source
• Dual Latent Memory in Vision-Language-Action Models for Robotic Manipulation score 10
  入选 HF Daily Papers; HF 热度: 50 upvotes (+4); 有代码实现; 关键词(2): reasoning, vision-language
• Infinite Worlds with Versatile Interactions score 10
  入选 HF Daily Papers; HF 热度: 26 upvotes (+4); 有代码实现; 关键词(5): lightweight, deployment, real-time, pretraining, agentic

Also Worth Noting

• GeoProp: Grounding Robot State in Vision for Generalist Manipulation score 4
  机构: Alibaba; 关键词(2): lightweight, embodied
• Creativity from Friction: Human-AI Interaction for Exploratory Structural Design score 4
  关键词(1): vision-language; 顶会接收: ICML