Pruning Context Can Cost More Than It Saves

Featured

• TokenPilot: Cache-Efficient Context Management for LLM Agents score 9
  入选 HF Daily Papers; HF 热度: 14 upvotes (+3); 有代码实现; 关键词(1): pruning

Also Worth Noting

• UniDDT: Unifying Multimodal Understanding and Generation with Decoupled Diffusion Transformer score 5
  入选 HF Daily Papers; HF 热度: 9 upvotes (+2)
• VinQA: Visual Elements Interleaved Long-form Answer Generation for Real-World Multimodal Document QA score 4
  关键词(1): fine-tuning; 顶会接收: CVPR
• The Quality-Utility Paradox: Why High-Reward Data Impairs Small Model Mathematical Reasoning score 4
  关键词(2): distillation, reasoning; 顶会接收: ICML
• Learning Interface Breakup: A Geometry-Conditioned Latent Surrogate for Spray Formation score 4
  关键词(1): lightweight; 顶会接收: ICML
• Creative Collision: Directorial Persona Steering and Competition in Large Language Models score 3
  顶会接收: ICML
• Simulation-Augmented Multi-Step Split Conformal Prediction for Aggregated Forecasts score 3
  顶会接收: ICML