arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2511.05850cs.IRcs.AI
收藏

上下文极限下的检索质量

Retrieval Quality at Context Limit

Google LLC Mountain View, U.S.A.(谷歌公司山景城美国分部)

Max McKinnon

AI总结:

本文发现 Gemini 2.5 Flash 在长上下文事实型问答中不受“Lost in the Middle”效应影响,即使文档接近上下文极限也能保持高检索准确率,表明长上下文检索能力显著提升。

摘要:

大型语言模型(LLMs)从长上下文中回忆和检索信息的能力对许多现实应用至关重要。先前的工作(Liu et al., 2023)报告称,LLMs 在检索位于大型上下文中间部分的事实时,检索准确率会显著下降,这一现象被称为“Lost in the Middle”(LITM)。我们发现模型 Gemini 2.5 Flash 能够以很高的准确率回答“大海捞针”(needle-in-a-haystack)式问题,无论文档位置如何,包括当文档几乎处于输入上下文极限时。我们的结果表明,在 Gemini 2.5 Flash 中,对于简单的事实型问答(factoid Q&A),“Lost in the Middle”效应并不存在,这表明长上下文检索能力有了显著改进。

英文摘要:

The ability of large language models (LLMs) to recall and retrieve information from long contexts is critical for many real-world applications. Prior work (Liu et al., 2023) reported that LLMs suffer significant drops in retrieval accuracy for facts placed in the middle of large contexts, an effect known as "Lost in the Middle" (LITM). We find the model Gemini 2.5 Flash can answer needle-in-a-haystack questions with great accuracy regardless of document position including when the document is nearly at the input context limit. Our results suggest that the "Lost in the Middle" effect is not present for simple factoid Q\&A in Gemini 2.5 Flash, indicating substantial improvements in long-context retrieval.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑