NCP Nearly Halves Training Tokens as T1 Reaches 64%

Featured

• Mi-Ripple: Restoring Images Degraded by Iterative AI Editing score 9
  入选 HF Daily Papers; HF 热度: 49 upvotes (+4); 有代码实现
• NCP-ArchPreview Technical Report: Moving towards Latent Space Language Models through Next Concept Prediction score 8
  入选 HF Daily Papers; HF 热度: 304 upvotes (+4); 关键词(2): lightweight, pretraining
• SenseNova-U1.5: Towards Native Unified Visual Intelligence score 8
  入选 HF Daily Papers; HF 热度: 248 upvotes (+4); 关键词(4): distillation, fine-tuning, post-training, open-source
• T1: Terminal Agent Reinforcement Learning for Long-Horizon Tasks score 8
  入选 HF Daily Papers; HF 热度: 51 upvotes (+4); 关键词(2): MoE, coding
• FreeFlow: A Bias-free Hierarchical Transformer for Optical Flow Estimation score 9
  入选 HF Daily Papers; HF 热度: 34 upvotes (+4); 有代码实现
• Negative Self-Distillation: Learning to Reason by Avoiding Flaws score 10
  入选 HF Daily Papers; HF 热度: 31 upvotes (+4); 有代码实现; 关键词(2): distillation, reasoning
• Generative Late-Interaction Embeddings For Visual Document Retrieval score 10
  入选 HF Daily Papers; HF 热度: 26 upvotes (+4); 有代码实现; 关键词(3): lightweight, compression, fine-tuning
• Recursive Code World Models: Building Complex Worlds through Recursive Scene Programs score 10
  入选 HF Daily Papers; HF 热度: 28 upvotes (+4); 有代码实现; 关键词(2): coding, vision-language
• An Open Recipe for IMO Gold: Training Nemotron for Olympiad Mathematics score 8
  入选 HF Daily Papers; HF 热度: 37 upvotes (+4); 关键词(2): fine-tuning, post-training
• UniH$^3$: Unifying Hierarchical Homogeneity and Heterogeneity for All-in-One Medical Image Restoration score 9
  入选 HF Daily Papers; HF 热度: 24 upvotes (+4); 有代码实现

Also Worth Noting

• DRG-MAPPO: Hierarchical Dynamic Role-Graph Multi-Agent Reinforcement Learning for Cooperative Air Combat score 7
  入选 HF Daily Papers; HF 热度: 22 upvotes (+4)
• Beyond Solver Verdicts: Generative Reward Models for Autoformalization score 8
  入选 HF Daily Papers; HF 热度: 28 upvotes (+4); 关键词(2): agentic, reasoning