跳到正文
研踪PaperTrail
← 研踪 PaperTrail

arXiv:2609.04910 · v2

Fast Gauss Sums via Flash Attention

首次提交 2026-09-04 17:11
版本更新 2026-11-19 16:39
雷达观测 2026-11-20 19:58
arXiv ↗PDF ↗

摘要

Gauss kernel sums are the computational core of maximum mean discrepancies (MMDs), kernel gradient flows, Stein variational gradient descent (SVGD), and many other kernel methods. At the same time, softmax attention has received an extraordinary amount of hardware-aware code engineering, culminating in FLASH attention. We show that Gauss kernel sums with arbitrary, signed weights can be evaluated via FLASH attention: two small input augmentations turn the normalized softmax reduction into the unnormalized Gauss sum, without writing a single line of custom GPU code. For feature dimension D>8D>8 in fp16, this approach beats compiled PyTorch code in speed and memory, and PyKeOps kernels in speed and accuracy at comparable or lower memory. Its memory scaling remains linear.

作者备注:6 pages 5 figures

AI 论文解读

基于摘要 · 非全文审阅

此版本尚未生成 AI 解读。解读由后台任务生成并按论文版本保存,登录后共享已有结果。

版本差异

本站首次收录为 v2,暂无更早版本可对照。

观测到的版本

相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。

v2 · 2026-11-19 16:39 · 最新观测版本

Fast Gauss Sums via Flash Attention

Gauss kernel sums are the computational core of maximum mean discrepancies (MMDs), kernel gradient flows, Stein variational gradient descent (SVGD), and many other kernel methods. At the same time, softmax attention has received an extraordinary amount of hardware-aware code engineering, culminating in FLASH attention. We show that Gauss kernel sums with arbitrary, signed weights can be evaluated via FLASH attention: two small input augmentations turn the normalized softmax reduction into the unnormalized Gauss sum, without writing a single line of custom GPU code. For feature dimension D>8D>8 in fp16, this approach beats compiled PyTorch code in speed and memory, and PyKeOps kernels in speed and accuracy at comparable or lower memory. Its memory scaling remains linear.

查看此版本解读 →查看此版本原文 ↗