13项技能把捏造检出率推至92%

重点关注

• AI4AI at Test-Time: Strong-to-Weak Capability Transfer via Harnesses score 8
  入选 HF Daily Papers；HF 热度: 107 upvotes (+4)；关键词(2): distillation, reasoning
• ToolHazard: Scaling Adversarial Environments for Security Evaluation and Alignment of LLM-based Agents score 8
  入选 HF Daily Papers；HF 热度: 8 upvotes (+2)；有代码实现；关键词(1): scaling
• StateFlow: Building, Evolving, and Accessing 3D World States for Previsualization score 8
  入选 HF Daily Papers；HF 热度: 25 upvotes (+4)；关键词(2): production, serving
• MBA: Multimodal Benchmark and Agents for Real-World Business Ideation score 8
  入选 HF Daily Papers；HF 热度: 6 upvotes (+2)；有代码实现；关键词(2): fine-tuning, agentic
• Spark-to-Paper: End-to-End Research Paper Generation as a Composable Skill score 10
  入选 HF Daily Papers；HF 热度: 274 upvotes (+4)；有代码实现；关键词(2): lightweight, coding

也值得关注

• From Synthesis to Removal: Physics-Grounded Reflection Simulation and Diffusion-Based Video Dereflection score 5
  入选 HF Daily Papers；HF 热度: 9 upvotes (+2)
• GeoFlow: Efficient Driving Video Generation via Geometry-Aligned Priors score 4
  关键词(2): latency, fine-tuning；顶会接收: ECCV
• AVA-Encoder: Towards Agent-Native Video Representation Learning score 6
  入选 HF Daily Papers；HF 热度: 5 upvotes (+2)；关键词(2): agentic, reasoning
• FunnelCausalNet: Funnel-aware Joint Conversion-Revenue Uplift for Multi-tier Coupon Allocation score 3
  机构: Alibaba
• Learning with Bilevel-Minimax Optimization for Efficient and Reliable Transfer Attacks score 3
  顶会接收: ECCV
• Dual Anchors, Do It Better: Hierarchical Group Merging for Zero-Shot Anomaly Detection score 3
  顶会接收: CVPR
• Earth observation embeddings are effective sub-grid descriptors for probabilistic weather downscaling score 3
  机构: Cambridge