跳到正文
Chelsea Finn· @chelseabfinn · X·· 4 天前AI 评分45
AI 导读

我们如何扩展LLM的上下文? 智能体搜索可以扩展到任意大量的信息,但也容易漏掉相关信息。 核心思路💡:为下游效用优化信息的组织方式。即使在OOD领域也能带来30%以上的性能提升 论文 + 代码:https://librarian-agents.github.io/

正文 · 原文

How do we scale up the context of LLMs?

Agentic search scales to arbitrarily large amounts of info, but can easily miss relevant info.

Idea💡: Optimize organization of info for downstream utility. Leads to 30+% gains even on OOD domains

Paper + code: https://librarian-agents.github.io/

来源:Chelsea Finn · x.com