11个系统未匹配未来概率，ES覆盖更多路径

重点关注

• UrbanGround: From Local Perception to Spatial Agency in a Real-Scale City score 10
  入选 HF Daily Papers；HF 热度: 71 upvotes (+4)；有代码实现；关键词(1): reasoning
• What Makes Good Agentic Data? An ACE Lens on Data Generation for LLM Agents score 8
  入选 HF Daily Papers；HF 热度: 59 upvotes (+4)；关键词(2): scaling, agentic
• Understanding Evolution Strategies for LLM Reasoning: Broader Reasoning Coverage than GRPO score 7
  入选 HF Daily Papers；HF 热度: 15 upvotes (+3)；关键词(3): GRPO, post-training, reasoning
• PAWBench: How Far Are We from Probabilistically Aligned World Modeling? score 9
  入选 HF Daily Papers；HF 热度: 82 upvotes (+4)；有代码实现

也值得关注

• Magpie: Real-Time World Renderer for Interactive Games score 6
  入选 HF Daily Papers；HF 热度: 7 upvotes (+2)；关键词(2): production, real-time
• A Single Suffix to Break Them All: Basin-Aware Jailbreaks for Merged Model Families score 4
  关键词(1): jailbreak；顶会接收: EMNLP
• Not Just Reason, Not Just Scan: Reinforcement Learning for Proactive Scientific Error Verification over Academic Paper score 4
  关键词(1): reasoning；顶会接收: EMNLP
• Do LLMs Understand Personality? Rethinking Persona Fidelity Evaluation through Structured Behavioral Inference score 4
  关键词(1): reasoning；顶会接收: EMNLP
• FOCUS & RePAIR: Mitigating Text Degeneration via Token-Level Guidance for Pruned Large Language Models score 4
  关键词(3): distillation, pruning, fine-tuning；顶会接收: ICML
• RuleWeaver: Benchmarking Rule-Centered Scenario Reasoning for Large Language Models score 4
  关键词(1): reasoning；顶会接收: EMNLP
• MedFG-VQA: Low-Frequency Memory and Graph Attention for Lightweight Medical VQA score 4
  关键词(3): lightweight, deployment, vision-language；顶会接收: CVPR
• From Reasoning to Pixels: Grounded Medical Multimodal LLMs for VQA and Segmentation score 4
  关键词(1): reasoning；顶会接收: ECCV
• Self-OPD: On-Policy Distillation for Flow Matching Models without Teacher score 8
  入选 HF Daily Papers；HF 热度: 67 upvotes (+4)；关键词(1): distillation
• RubricRM: Generative Reward Modeling via Dynamic Rubrics for Image Generation and Editing score 4
  关键词(3): fine-tuning, GRPO, text-to-image；顶会接收: EMNLP
• Research Design Tracking and Assessment for the Social Sciences score 4
  关键词(1): RAG；顶会接收: EMNLP
• Prediction of Prediction (PoP): Inter-Layer Activation Fusion for Single-Pass Hallucination Detection in Large Language Models score 4
  机构: Mistral；关键词(2): deployment, latency
• Ancient-Bench: A Comprehensive Multi-millennial, Multi-medium, and Multi-script Benchmark for Ancient Chinese Artifact Text Recognition score 4
  关键词(1): vision-language；顶会接收: EMNLP
• BALMS: Benchmarking Agentic LLMs for Longitudinal Mental Health Sensing score 4
  关键词(3): scaling, agentic, reasoning；顶会接收: EMNLP
• MM-Spectrum: Multimodal Multi-spectral Molecular Structural Elucidation with a Stable MoE Framework score 4
  关键词(1): MoE；顶会接收: ICML