全部动态

AI 相关资讯全量信息流 · 6327 条
8月27日
论文09:54
One Symptom, Three Levers: A Critical Review of On-Policy Self-Distillation

This review delves into the challenges of On-Policy Self-Distillation, highlighting the 'collapse' issue and offering insights into its causes. It's a must-read for those interested in understanding the nuances of model training with privileged information.

arXiv cs.AI@Justin Robert, Raheel Qader原文
论文09:46
Formal, Executable and Explainable Runtime Monitoring of Spoken Air Traffic Control Operational Procedures

这篇论文提出了一种新的空中交通管制程序运行时监控框架,通过结合无线电通信、监视数据和机载观察,能够有效识别程序偏差,对于提高空中交通安全具有重要意义。与传统的监控方法相比,该方法在真实交通和合成情况下的表现都十分出色,值得深入研究。

arXiv cs.AI@Roberto Luvini 等 4 人原文
论文09:20
Where vs What: Decomposing Structural and Content Failures in LLM-Generated Structured Outputs

Read this paper if you're interested in understanding how to measure and improve the structural accuracy of LLM-generated structured outputs. It introduces a new framework and a method that could lead to better AI applications.

arXiv: DeepSeek@Yiwei Zhang 等 4 人原文
论文Agent 与开发者多源确认09:18
EVOMAL: 自进化的编码智能体中的自毒化问题

研究揭示了自进化的编码智能体中的自毒化问题,并提出了counter-prompt防御措施,有效降低了恶意技能的传播。对于关注AI安全和智能体研究的读者来说,这是一篇不容错过的论文。

事件专题
arXiv cs.AI@Xiaodong Wu 等 8 人3 个信源在谈原文
8月26日

仅展示最近 2000 条内容,更早的内容请查阅 AI 日报存档