FID当loss一步生成达0.72

重点关注

• Heterogeneous Scientific Foundation Model Collaboration score 10
  入选 HF Daily Papers；HF 热度: 186 upvotes (+4)；有代码实现；关键词(2): agentic, reasoning
• Representation Fr\'echet Loss for Visual Generation score 9
  入选 HF Daily Papers；HF 热度: 13 upvotes (+3)；有代码实现；关键词(2): distillation, post-training
• Visual Generation in the New Era: An Evolution from Atomic Mapping to Agentic World Modeling score 8
  入选 HF Daily Papers；HF 热度: 74 upvotes (+4)；关键词(6): distillation, post-training, agentic, reasoning, synthetic data
• InteractWeb-Bench: Can Multimodal Agent Escape Blind Execution in Interactive Website Generation? score 8
  入选 HF Daily Papers；HF 热度: 9 upvotes (+2)；有代码实现；关键词(1): coding
• World2Minecraft: Occupancy-Driven Simulated Scenes Construction score 7
  入选 HF Daily Papers；HF 热度: 2 upvotes (+1)；有代码实现；关键词(2): vision-language, embodied
• MoCapAnything V2: End-to-End Motion Capture for Arbitrary Skeletons score 8
  入选 HF Daily Papers；HF 热度: 6 upvotes (+2)；有代码实现；关键词(1): reasoning
• Synthetic Computers at Scale for Long-Horizon Productivity Simulation score 7
  入选 HF Daily Papers；HF 热度: 12 upvotes (+3)；关键词(2): agentic, synthetic data

也值得关注

• Intern-Atlas: A Methodological Evolution Graph as Research Infrastructure for AI Scientists score 7
  入选 HF Daily Papers；HF 热度: 22 upvotes (+4)
• ScaleBox: Enabling High-Fidelity and Scalable Code Verification for Large Language Models score 4
  关键词(2): throughput, coding；顶会接收: ACL
• Decoding Scientific Experimental Images: The SPUR Benchmark for Perception, Understanding, and Reasoning score 4
  关键词(1): reasoning；顶会接收: ACL
• Taming Noise-Induced Prototype Degradation for Privacy-Preserving Personalized Federated Fine-Tuning score 4
  关键词(2): distillation, fine-tuning；顶会接收: CVPR
• Latent-GRPO: Group Relative Policy Optimization for Latent Reasoning score 4
  机构: Huawei；关键词(2): GRPO, reasoning
• FedHarmony: Harmonizing Heterogeneous Label Correlations in Federated Multi-Label Learning score 4
  关键词(1): serving；顶会接收: CVPR
• PhyCo: Learning Controllable Physical Priors for Generative Motion score 6
  入选 HF Daily Papers；HF 热度: 7 upvotes (+2)；关键词(2): fine-tuning, vision-language
• AEGIS: A Holistic Benchmark for Evaluating Forensic Analysis of AI-Generated Academic Images score 4
  关键词(1): reasoning；顶会接收: ACL
• Iterative Definition Refinement for Zero-Shot Classification via LLM-Based Semantic Prototype Optimization score 3
  顶会接收: CVPR
• Investigating More Explainable and Partition-Free Compositionality Estimation for LLMs: A Rule-Generation Perspective score 3
  顶会接收: ACL
• HealthBench Professional: Evaluating Large Language Models on Real Clinician Chats score 3
  机构: OpenAI
• Debiasing Reward Models via Causally Motivated Inference-Time Intervention score 3
  顶会接收: ACL
• REVIVE 3D: Refinement via Encoded Voluminous Inflated prior for Volume Enhancement score 3
  顶会接收: CVPR
• APPSI-139: A Parallel Corpus of English Application Privacy Policy Summarization and Interpretation score 3
  顶会接收: ACL
• SECOS: Semantic Capture for Rigorous Classification in Open-World Semi-Supervised Learning score 3
  顶会接收: CVPR