视觉模型开始重新设计自己的输出方式

重点关注

• SpatialBench: Is Your Spatial Foundation Model an All-Round Player? score 10
  入选 HF Daily Papers；HF 热度: 63 upvotes (+4)；有代码实现；关键词(2): scaling, embodied
• Gemini Embedding 2: A Native Multimodal Embedding Model from Gemini score 7
  入选 HF Daily Papers；HF 热度: 15 upvotes (+3)；关键词(1): RAG
• LocateAnything: Fast and High-Quality Vision-Language Grounding with Parallel Box Decoding score 8
  入选 HF Daily Papers；HF 热度: 114 upvotes (+4)；关键词(2): throughput, vision-language
• JLT: Clean-Latent Prediction in Latent Diffusion Transformers score 10
  入选 HF Daily Papers；HF 热度: 26 upvotes (+4)；有代码实现；关键词(1): compression
• MRT: Masked Region Transformer for Layered Image Generation and Editing at Scale score 6
  入选 HF Daily Papers；HF 热度: 5 upvotes (+2)；关键词(2): distillation, real-time

也值得关注

• Verus-SpecGym: An Agentic Environment for Evaluating Specification Autoformalization score 7
  入选 HF Daily Papers；HF 热度: 3 upvotes (+1)；有代码实现；关键词(3): edge, agentic, coding
• PlayClass: Automated Play Behaviour Classification in Poultry score 3
  机构: Imperial College