Move to See: Top Model Reaches the Target Just 12% of the Time

Featured

• Where to Look: Can Foundation Models Reach a Target Viewpoint Through Active Exploration? score 10
  入选 HF Daily Papers; HF 热度: 20 upvotes (+4); 有代码实现; 关键词(4): GRPO, post-training, embodied, open-source
• SkillAdaptor: Self-Adapting Skills for LLM Agents from Trajectories score 7
  入选 HF Daily Papers; HF 热度: 20 upvotes (+4)
• HakushoBench: A Japanese Chart and Table VQA Benchmark from Governmental White Papers score 5
  入选 HF Daily Papers; HF 热度: 4 upvotes (+1); 关键词(1): vision-language
• 3DCodeBench: Benchmarking Agentic Procedural 3D Modeling Via Code score 5
  入选 HF Daily Papers; HF 热度: 3 upvotes (+1); 关键词(5): scaling, agentic, coding, reasoning, vision-language

Also Worth Noting

• LongAttnComp: Cross-Family Context Compression for Long-Context Reasoning score 5
  入选 HF Daily Papers; HF 热度: 3 upvotes (+1); 关键词(4): lightweight, compression, fine-tuning, reasoning
• Practical and Optimal Algorithm for Linear Contextual Bandits with Rare Parameter Updates score 3
  顶会接收: ICML
• Digging Up Citations: FOSSIL, a Dataset and Workflow for Reference Extraction in Law and the Humanities score 3
  机构: Google