返回全部动态
TREC 2025 百万大语言模型赛道概览
原标题：Overview of the TREC 2025 Million Large Language Models track
AI 摘要
TREC 2025 百万大语言模型赛道（Million LLMs Track）提出一种新的检索范式：不再检索文档，而是检索专家模型。该赛道提供超过一千个 LLM 的标识符，以及由查询、答案和 token 级对数概率组成的发现数据集，要求参赛系统根据模型的可观察行为推断其专长，并对未见测试查询按预期表现对模型排序。这是面向智能体 AI 中专家模型检索的首个大规模基准。
以上摘要由 AI 生成，可能存在误差。事实请以原文为准。
正文节选
Overview of the TREC 2025 Million Large Language Models track Abstract Agentic AI envisions ecosystems of intelligent agents collaboratively solving complex tasks with minimal human intervention. In such ecosystems, each agent possesses specialized expertise, making effective expert selection central to overall system performance. While most current approaches assume a small number of well-documented models, real-world expertise is far more diverse and cannot be adequately captured through stati
发布时间：2026-09-29 12:00
抓取时间：2026-09-29 13:58
来源机构：arXiv