NVIDIA五模态压进一套权重

重点关注

• World Models Meet Language Models: On the Complementarity of Concrete and Abstract Reasoning score 10
  入选 HF Daily Papers；HF 热度: 29 upvotes (+4)；有代码实现；关键词(2): distillation, reasoning
• Cosmos 3: Omnimodal World Models for Physical AI score 10
  入选 HF Daily Papers；HF 热度: 83 upvotes (+4)；有代码实现；关键词(5): deployment, vision-language, text-to-image, embodied, open-source
• KVarN: Variance-Normalized KV-Cache Quantization Mitigates Error Accumulation in Reasoning Tasks score 10
  入选 HF Daily Papers；HF 热度: 50 upvotes (+4)；有代码实现；关键词(3): scaling, quantization, reasoning
• Language Models Need Sleep: Learning to Self-Modify and Consolidate Memories score 8
  入选 HF Daily Papers；HF 热度: 23 upvotes (+4)；关键词(2): distillation, synthetic data
• NVIDIA OmniDreams: Real-Time Generative World Model for Closed-Loop Autonomous Vehicle Simulation score 11
  机构: NVIDIA；入选 HF Daily Papers；HF 热度: 20 upvotes (+4)；关键词(2): real-time, post-training
• Benchmarking Visual State Tracking in Multimodal Video Understanding score 10
  入选 HF Daily Papers；HF 热度: 21 upvotes (+4)；有代码实现；关键词(2): agentic, coding
• PaddleOCR-VL-1.6: Expanding the Frontier of Document Parsing with Under-Optimized Region Refinement and Progressive Post-Training score 9
  入选 HF Daily Papers；HF 热度: 13 upvotes (+3)；有代码实现；关键词(1): post-training
• Value-Aware Stochastic KV Cache Eviction for Reasoning Models score 8
  入选 HF Daily Papers；HF 热度: 8 upvotes (+2)；有代码实现；关键词(2): compression, reasoning
• Small RL Controller, Large Language Model: RL-Guided Adaptive Sampling for Test-Time Scaling score 7
  入选 HF Daily Papers；HF 热度: 13 upvotes (+3)；关键词(4): scaling, lightweight, latency, reasoning
• Eliciting Complex Spatial Reasoning in MLLMs through Wide-Baseline Matching score 9
  入选 HF Daily Papers；HF 热度: 15 upvotes (+3)；有代码实现；关键词(1): reasoning

也值得关注

• Conditional Hypothesis Generation for LLM-Based Text Analysis with Researcher-Specified Covariates score 5
  入选 HF Daily Papers；HF 热度: 5 upvotes (+2)
• AUDITFLOW: Executable Symbolic Environments for Structured Financial Reporting Verification score 5
  入选 HF Daily Papers；HF 热度: 6 upvotes (+2)
• SynCred-Bench: Benchmarking Synthetic Credibility in AI-Generated Visual Misinformation score 5
  入选 HF Daily Papers；HF 热度: 2 upvotes (+1)；关键词(1): open-source
• Economy of Minds: Emerging Multi-Agent Intelligence with Economic Interactions score 8
  入选 HF Daily Papers；HF 热度: 8 upvotes (+2)；有代码实现；关键词(2): agentic, reasoning
• Evaluating LLMs' Effectiveness on Real-World Consumer Device Repair Questions score 4
  机构: University of Toronto；关键词(1): reasoning
• OVO-S-Bench: A Hierarchical Benchmark for Streaming Spatial Intelligence in Multimodal LLMs score 10
  入选 HF Daily Papers；HF 热度: 30 upvotes (+4)；有代码实现；关键词(4): serving, reasoning, robotics, open-source