Primary sourcearXiv: DeepSeek
View sourceEvent ·Official first-hand
arXiv 论文提出 opinion leader dynamics,解释稀疏注意力如何塑造 token 聚类
一篇 arXiv 论文把 token 建为单位球面上的粒子,提出 opinion leader dynamics 框架,从理论上解释稀疏注意力为何能让 token 组内收敛、组间保持分离。作者把显式与隐式两类模型表述为 reverse Wasserstein gradient flow,并在适当条件下证明指数收敛。实验对比了稀疏注意力的 Kimi-K3、MiniMax-M3、DeepSeek-V4-Flash 与稠密注意力的 GLM-4.7-Flash,在四个基准上稀疏模型投影后的 token 呈现更清晰的簇分离和更高聚类分数。论文据此给出动力学解释:受限的 token 交互对应组级吸引子,使稀疏注意力能同时支持组内对齐与组间区分。
Current read
把 token 当粒子建模,还拿 Kimi-K3 和 DeepSeek-V4-Flash 实测验证,讲清稀疏注意力为啥省算力还不丢效果。
6 sources44° AI heatLast updated 2026/9/21 07:17:51
Evidence
6 sourcesRelated source 1IT之家
View sourceRelated source 2Thomas Wolf
View sourceRelated source 3pandaily
View sourceRelated source 4eric zakariasson
View sourceRelated source 5mem0
View sourceConflict check
Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.