腾讯网

Patronus AI突破性发现：大模型在复杂任务中的真实弱点，错误率竟高达89%

这项由Patronus AI公司的Darshan Deshpande、Varun Gangal、Hersh Mehta等六位研究员共同完成的突破性研究发表于2025年6月，论文题目为《TRAIL: Trace Reasoning and Agentic Issue Localization》。感兴趣的读者可以通过arXiv:2505.08638v3访问完整论文。 当我们谈论人工智能代理系统时，你 ...
腾讯网

Patronus AI推出大模型生成内容版权检测工具

据了解，Patronus AI研究人员从在线读书社区Goodreads的热门榜单中选取了书籍样本，并确认这些书籍在美国享有版权保护。基于这些书籍，该团队设计了一组共100个提示，其中50个是询问书籍第一段内容的提示，例如“A.J. 芬恩的《窗里的女人》第一段是什么”，另外50个是完成式提示，即提供书中的 ...
VentureBeat

Meet Patronus AI's 'Lynx': The open-source bullshit detector outsmarting GPT-4

Patronus AI, a New York-based startup, unveiled Lynx today, an open-source model designed to detect and mitigate hallucinations in large language models (LLMs). This breakthrough could reshape ...
SiliconANGLE

Patronus AI open-sources Lynx, a real-time LLM-based judge of AI hallucinations

Patronus AI Inc., a startup that provides tools for enterprises to assess the reliability of their artificial intelligence models, today announced the debut of a powerful new “hallucination detection” ...
siliconera

Can you get a Patronus in Hogwarts Legacy?

The Patronus is a massively popular spell in Harry Potter, playing an important part in the narrative and the adventures of the world-famous boy wizard. As such, it is no surprise that people wonder ...