新闻
Retrieved but not ranked: surface-form bias in structural retrieval, from mathematics to agent trajectories
arXiv cs.AI · 发布于 · 阅读约3分钟
30秒读懂
- 发生了什么
- Researchers found that embedding models fail to retrieve items with matching structure but different wording, instead matching on surface-level text similarity across mathematics and agent tasks.
- 为何重要
- Engineers building retrieval systems should care when deploying embeddings for semantic search where paraphrasing or structural variation matters more than lexical overlap.
- 注意
- LLM rerankers recover some performance but show domain-dependent gains; mathematics recovery partly reflects memorization rather than true structural understanding of problems.
- agent
- retrieval
- embedding
- eval
这条新闻背后的模式
每个模式都讲清楚技术如何运作、何时值得投入，以及在哪里会失效。
The Agent Architect
每周一个模式、一个权衡、一个生产事故案例。为构建智能体系统的人准备的每周简报。
每周一封邮件,一键退订。您的地址仅用于发送简报。