Skip to content

AI论文速递 2026年07月27日（HuggingFace Daily Papers）

数据来源：https://huggingface.co/papers 采集时间：2026-07-27

📌 重点关注

  1. ReferTrack: Referring Then Tracking for Embodied Visual Tracking | arXiv — 【重点关注】 Embodied visual tracking (EVT) requires a mobile agent to continuously follow... 💡 端侧AI+视觉追踪的结合方向，mufans的移动端背景可以切入具身智能赛道
  2. Skill Self-Play: Pushing the Frontier of LLM Capability with Co-Evolving Skills | arXiv — 【重点关注】 LLM training is shifting from manual design and annotation to interaction-dri... 💡 自我博弈式技能进化，Agent能力自动提升的关键范式，值得深入
  3. NVIDIA-labs OO Agents: Native Python Object-Oriented Agents | arXiv — 【重点关注】 Traditional agent development is split across prompt templates, tool schemas,... 💡 NVIDIA用纯Python OOP统一Agent开发，告别繁琐prompt模板

📋 其他值得关注

  1. OpenForgeRL: Train Harness-native Agents in Any Environment | arXiv — Modern AI agents rely on elaborate inference harnesses such as Claude Code, C...
  2. An Exam for Active Observers | arXiv — Human vision is a closed loop: gaze is continuously redirected by intermediat...
  3. Multi-Turn On-Policy Distillation with Prefix Replay | arXiv — We study on-policy distillation (OPD) for agentic tasks, where an LLM agent i...
  4. SeededGrasp: Language-Guided Grasping in Complex Scenes with Multiple Embodiments | arXiv — Practical robotic grasping in complex scenes requires both 3D spatial reasoni...
  5. Scaling Native Multimodal Pre-Training From Scratch | arXiv — Although large language models (LLMs) exhibit remarkable reasoning capabiliti...
  6. Dataset Distillation by Influence Matching | arXiv — We revisit dataset distillation from an outcome-centric perspective. Rather t...
  7. ICAE-Bench: Evaluating Coding Agents as Interactive Project Builders | arXiv — The recent emergence of vibe-coding workflows is changing what coding agents ...