English(EN) Remember to Forget: Gated Adaptive Positional Encoding

新的GAPE方法增强了LLM的长上下文泛化能力

作者 PulseAugur 编辑部 · [1 个来源] · 2026-10-09 04:00

研究人员开发了一种新颖的方法，称为门控自适应位置编码（GAPE），以提高大型语言模型（LLM）的长上下文处理能力。GAPE解决了当序列超出训练长度时旋转位置编码（RoPE）存在的问题，该问题可能导致性能下降。通过向注意力对数引入感知内容的偏差，GAPE选择性地抑制不相关上下文，同时保留重要标记，使模型能够泛化到比其训练数据长八倍的上下文长度。 AI

影响增强了LLM处理和回忆扩展上下文信息的能力，可能提高需要长距离理解的任务的性能。

排序理由详细介绍一种改进LLM长上下文能力的新方法的学术论文。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.LG 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: 0 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 详细介绍一种改进LLM长上下文能力的新方法的学术论文。[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, model release; 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: 14 days old; Aged out of breaking-news scoring windows; ranking reflects the durable signal from the full source set.

完整方法见我们的编辑标准。

报道来源 [1]

arXiv cs.LG TIER_1 English(EN) · Riccardo Ali, Alessio Borgi, Mario Severino, Pietro Li\`o, Christopher Irwin · 2026-10-09 04:00

记住遗忘：门控自适应位置编码

arXiv:2605.10414v2 Announce Type: replace Abstract: Rotary Positional Encoding (RoPE) is widely used in modern large language models. However, when sequences are extended beyond the range seen during training, rotary phases can enter out-of-distribution regimes, leading to spurio…

本文如何被排名

报道来源 [1]

记住遗忘：门控自适应位置编码

相关实体

相关话题