English(EN) Sub-Token Routing in LoRA for Adaptation and Query-Aware KV Compression

子令牌路由在新的研究中提高了Transformer的效率和KV压缩能力

作者 PulseAugur 编辑部 · [1 个来源] · 2026-05-07 04:00

研究人员引入了子令牌路由作为一种新颖的方法来提高Transformer的效率，它提供了一种比现有技术更精细的压缩方法。该方法侧重于在令牌表示内部进行路由，并探索了查询无关和查询感知的设置。实验表明，将子令牌路由与令牌级选择相结合，可以在保持高任务准确性的同时实现显著的KV压缩。 AI

影响引入了一种新的压缩技术，有望实现内存需求更低、效率更高的大型语言模型。

排序理由这是一篇发表在arXiv上的研究论文，详细介绍了一种新的Transformer效率技术方法。[lever_c_demoted from research: ic=1 ai=1.0]

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; 这是一篇发表在arXiv上的研究论文，详细介绍了一种新的Transformer效率技术方法。[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, infra; 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: 149 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。

arXiv cs.LG TIER_1 English(EN) · Wei Jiang, Wei Wang · 2026-05-07 04:00

LoRA中的子Token路由用于适配和查询感知KV压缩

arXiv:2604.21335v2 Announce Type: replace Abstract: Sub-token routing provides a finer compression axis for transformer efficiency than the coarse units used in most prior work, such as tokens, pages, heads, or layers. In this paper, we study routing within a token representation…