English(EN) Training-Induced Escape from Token Clustering in a Mean-Field Formulation of Transformers

新理论解释了 Transformer 训练如何影响 Token 聚类

作者 PulseAugur 编辑部 · [1 个来源] · 2026-10-19 04:00

Transformer 动力学的新理论公式揭示了训练过程如何在推理过程中影响 Token 聚类。研究人员发现，即使是单次梯度下降更新也会导致后续层中的 Token 分布逃逸聚类。这项工作建立了一个训练感知的均场理论，将训练和推理动力学结合起来，以更好地理解 Transformer 的行为。 AI

影响为理解训练如何影响 Transformer 推理动力学提供了理论框架。

排序理由该集群包含一篇关于 Transformer 模型理论方面的学术论文。[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: 9 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 该集群包含一篇关于 Transformer 模型理论方面的学术论文。[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, 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: Same-day; Cluster formed today. Ranking reflects the current source set at time of score.

完整方法见我们的编辑标准。

arXiv cs.LG TIER_1 English(EN) · Noboru Isobe, Daisuke Inoue, Masaaki Imaizumi · 2026-10-19 04:00

Transformer均场公式中训练诱导的Token聚类逃逸

arXiv:2605.07772v2 Announce Type: replace Abstract: Transformers perform inference by iteratively transforming token representations across layers. This layerwise computation has been studied empirically, and recent mean-field theories of Transformer dynamics explain how attentio…