AI论文速递 2026年08月12日（HuggingFace Daily Papers）

数据来源：https://huggingface.co/papers 采集时间：2026-08-12

📌 重点关注

1. SPIEval: Evaluating Large Language Models as Mobile Assistants over Scattered Personal Information | arXiv — 【重点关注】 Large language models (LLMs) are increasingly deployed as mobile assistants, ... 💡 LLM手机助手评估基准，移动端+AI转型的核心方向
2. A^2E : An End-to-End Agent Auditing Engine | arXiv — 【重点关注】 With the rapid advancement of large language models (LLMs), harnesses have be... 💡 端到端Agent审计引擎，是Agent可靠落地的关键基础设施
3. Multi-Agent Forensic Reasoning for Generalizable Deepfake Video Detection | arXiv — 【重点关注】 The malicious use of generative artificial intelligence to create highly real... 💡 多Agent协作做深度伪造检测，多Agent架构实战参考

📋 其他值得关注

1. Motif 3: Technical Report | arXiv — We introduce Motif 3, a decoder-only Mixture-of-Experts language model with 3...
2. Stealing Reasoning Traces from Proprietary LLM APIs | arXiv — Leading large language model providers now conceal their models' step-by-step...
3. Agent Memory Distillation: Empowering Small LLM Agents with Hierarchical Teacher Memory | arXiv — Memory systems have shown promise for improving agent performance, but their ...
4. RoMeRL: Balancing Feedback Coverage and the Memory-Reward Trap in Self-Evolving Agent Memory via Reduced-Order Utility States | arXiv — Learning-based memory systems for self-evolving LLM agents face two tightly c...
5. On-Policy Self-Distillation without Any Supervision | arXiv — On-policy (Self-)Distillation (OPD / OPSD) has shown strong potential for pos...
6. The Next Screenshot Knows: Gated Hindsight Distillation for Mobile GUI Agents | arXiv — GUI agents are commonly trained offline from successful interaction trajector...
7. OasisKV: Scaling In-Decode KV Cache Beyond HBM with Lookahead Sparse Prefetching | arXiv — Large language model (LLM) inference serving is increasingly constrained by m...