English(EN) When Attention Closes: How LLMs Lose the Thread in Multi-Turn Interaction

新框架解释大型语言模型在多轮对话中的上下文丢失问题

作者 PulseAugur 编辑部 · [1 个来源] · 2026-10-07 04:00

研究人员提出了一个新框架，用以解释大型语言模型（LLMs）为何在冗长、多轮的对话中难以保持上下文。他们的“通道-转换模型”表明，虽然对关键指令的注意力可能会减弱，但信息可以在模型内部的残余表示中持续存在。为了量化这一点，他们引入了目标可达性比率（GAR），并用它来分析各种架构，发现注意力消散的点在不同模型之间存在显著差异。 AI

影响为理解和潜在缓解大型语言模型中的上下文丢失问题提供了一个新框架，这对于开发更强大的对话式人工智能至关重要。

排序理由学术论文，详细介绍了大型语言模型行为的新机制解释和诊断工具。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.AI 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。我们如何撰写摘要 →

本文如何被排名

Signal score: 20 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 学术论文，详细介绍了大型语言模型行为的新机制解释和诊断工具。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration: Single-source cluster; Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics: paper, other; Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance: High; Clearly on-topic for AI-industry coverage.
Story freshness: Breaking (< 6h); Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

完整方法见我们的编辑标准。

报道来源 [1]

arXiv cs.AI TIER_1 English(EN) · Vardhan Dongre, Joseph Hsieh, Viet Dac Lai, Seunghyun Yoon, Trung Bui, Dilek Hakkani-T\"ur · 2026-10-07 04:00

当注意力关闭时：大型语言模型在多轮交互中如何失去线索

arXiv:2605.12922v2 Announce Type: replace Abstract: Large language models can follow complex instructions in a single turn, yet over long multi-turn interactions they often lose the thread of instructions, persona, and rules. This degradation has been measured behaviorally but no…

本文如何被排名

报道来源 [1]

当注意力关闭时：大型语言模型在多轮交互中如何失去线索

相关实体

相关话题