AI论文速递 2026年05月28日（HuggingFace Daily Papers）¶
数据来源：https://huggingface.co/papers 采集时间：2026-05-28
📌 重点关注¶
- Learning to Act under Noise: Enhancing Agent Robustness via Noisy Environments | arXiv — 【重点关注】 Recent advances in large language models (LLMs) have facilitated the widespre...
💡 AI实用稳定性优化新思路 2. LLaVA-OneVision-2: Towards Next-Generation Perceptual Intelligence | arXiv — 【重点关注】 We introduce LLaVA-OneVision-2 (LLaVA-OV-2), the most capable vision-language...
💡 多模态感知智能的重大突破 3. Agent Explorative Policy Optimization for Multimodal Agentic Reasoning | arXiv — 【重点关注】 Vision-language models with extended reasoning succeed on complex problems, b...
💡 复杂推理场景的优化方案
📋 其他值得关注¶
- Efficient Agentic Reinforcement Learning with On-Policy Intrinsic Knowledge Boundary Enhancement | arXiv — Agentic reinforcement learning (RL) has proven effective for training LLM-bas...
- MobileMoE: Scaling On-Device Mixture of Experts | arXiv — Mixture-of-Experts (MoE) has become the de facto architecture for hundred-bil...
- VitaBench 2.0: Evaluating Personalized and Proactive Agents in Long-Term User Interactions | arXiv — Large language models (LLMs) have evolved into interactive agents that collab...
- QUACK: Questioning, Understanding, and Auditing Communicated Knowledge in Multimodal Social Deduction Agents | arXiv — Social deduction games have become a popular testbed for probing reasoning, d...
- Learn from Weaknesses: Automated Domain Specialization for Small Computer-Use Agents | arXiv — Computer-use agents (CUAs) have recently made substantial progress, but deplo...
- ResearchMath-14K: Scaling Research-Level Mathematics via Agents | arXiv — The frontier of mathematics is defined by problems whose solutions are not ye...
- From Pixels to Words -- Towards Native One-Vision Models at Scale | arXiv — Current vision-language models (VLMs) typically stitch together separate imag...