English(EN) XMerge: Cross-Axis Selection and Reconstructive Layer Merging for LLM Depth Compression

XMerge方法在无需微调的情况下压缩LLM深度

作者 PulseAugur 编辑部 · [1 个来源] · 2026-09-03 04:00

研究人员开发了XMerge，一种新颖的训练后方法，旨在压缩大型语言模型（LLM）的深度，而无需任务特定的标签或端到端微调。该技术识别对模型输出影响最小的Transformer层，并重建相邻层以保持性能。XMerge在各种Llama和Qwen模型上均取得了优于现有方法的成果，尤其是在积极移除层的情况下，在降低困惑度和保持校准方面表现更佳。 AI

影响该方法可以通过减小LLM的尺寸而不会显著降低性能，从而实现更高效的LLM部署。

排序理由该集群包含一篇详细介绍LLM深度压缩新方法的论文。[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: 22 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 该集群包含一篇详细介绍LLM深度压缩新方法的论文。[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) · Jundong Hu, Shekar Ramachandran · 2026-09-03 04:00

XMerge：用于LLM深度压缩的跨轴选择和重构层合并

arXiv:2609.02083v1 Announce Type: cross Abstract: Removing complete transformer layers preserves a standard serving architecture, but existing depth-compression methods can lose substantial quality, and the loss varies unpredictably across models. We introduce XMerge, a post-trai…

本文如何被排名

报道来源 [1]

XMerge：用于LLM深度压缩的跨轴选择和重构层合并

相关实体

相关话题