用户改代码让Agent成功率降7.7个百分点

重点关注

• UEmbed: Unified Sparse and Dense Multimodal Embeddings score 10
  入选 HF Daily Papers；HF 热度: 43 upvotes (+4)；有代码实现；关键词(2): retrieval-augmented, agentic
• SWE-Touch: Benchmarking Coding Agents When Users Touch the Code score 10
  入选 HF Daily Papers；HF 热度: 23 upvotes (+4)；有代码实现；关键词(1): coding
• GradCuit: Credit-Assigned Gradient Flow Enables Robust and Interpretable Test-Time Latent Reasoning score 10
  入选 HF Daily Papers；HF 热度: 22 upvotes (+4)；有代码实现；关键词(2): scaling, reasoning
• WorldExam: Benchmarking World Models from Apparent Appearance to Inherent Reactivity score 9
  入选 HF Daily Papers；HF 热度: 30 upvotes (+4)；有代码实现
• GEOID-Flood: A Large-Scale Multi-Modal Benchmark Dataset for Flood Segmentation score 9
  入选 HF Daily Papers；有代码实现；关键词(1): finetuning；顶会接收: ECCV
• SKT: Skill-Use Training at Scale via Verified Synthetic Data Generation score 8
  入选 HF Daily Papers；HF 热度: 28 upvotes (+4)；关键词(3): scaling, fine-tuning, synthetic data
• ScrambleToolBench: Agents Search Exhaustively Even When Their Own Map Points to the Next Step score 9
  入选 HF Daily Papers；HF 热度: 10 upvotes (+3)；有代码实现；关键词(1): reasoning

也值得关注

• Sen-Cap: Sensor-Flexible and Noise-Resilient Human Motion Capture via LiDAR-Camera Integration score 4
  关键词(2): deployment, robotics；顶会接收: ECCV
• Aggregate-then-Calibrate for Human-centered Assessment with Theoretical Guarantees score 3
  顶会接收: ICLR