English(EN) DepthGraft: Structural Regularization through Hierarchical Cross-Layer KV Reconstruction

DepthGraft架构通过分层KV共享增强LLM性能

作者 PulseAugur 编辑部 · [1 个来源] · 2026-11-04 05:00

研究人员推出了一种新颖的分层KV共享架构DepthGraft，旨在通过解决深层信息稀释问题来提高大型语言模型的性能。该方法从跨步对齐的浅层-中层源对中重建深层KV表示，从而实现非局部梯度流并丰富KV内容子空间。在0.5B到16B参数的模型上进行的实验表明，下游性能得到了一致的提升，包括在16B MoE模型上MMLU-Pro的绝对增益达到7.0%。 AI

影响增强了LLM的训练效率和性能，有望以更低的计算成本实现更强大的模型。

排序理由这是一篇详细介绍大型语言模型新架构的研究论文。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.CL 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: 41 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 这是一篇详细介绍大型语言模型新架构的研究论文。[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: 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.CL TIER_1 English(EN) · Xiaohan Qin, Xiangdong Zhang, Yu Wang, Huaijin Wu, Zhuo Xia, Yebin Yang, Junchi Yan · 2026-11-04 05:00

DepthGraft：通过分层跨层KV重建实现结构正则化

arXiv:2611.03142v1 Announce Type: new Abstract: Scaling depth is a key route to stronger large language models, but Pre-LN architectures often suffer from depth-wise information dilution, limiting the effective use of deep layers. While cross-layer connectivity has proven effecti…

本文如何被排名

报道来源 [1]

DepthGraft：通过分层跨层KV重建实现结构正则化

相关实体

相关话题