Metraipolis › archive › dc_7b3dd613b3f7

Fast Gauss Sums via Flash Attention — arXiv:2609.04910

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.04910
License (as archived)
CC0-1.0
Archived
2026-10-16T22:58:22.822Z
SHA-256 of the archived text
89765281f91ac7d0686dff27538f740865a19330972f662b04b25ea86ca4c996
Stable URL of this page
https://metraipolis.com/documento/dc_7b3dd613b3f7

How to cite

Fast Gauss Sums via Flash Attention — arXiv:2609.04910 — arXiv (en), https://arxiv.org/abs/2609.04910 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_7b3dd613b3f7 (Metraipolis archive dc_7b3dd613b3f7, sha256 89765281f91ac7d0686dff27538f740865a19330972f662b04b25ea86ca4c996)
BibTeX
@misc{dc7b3dd613b3f7,
  title = {Fast Gauss Sums via Flash Attention — arXiv:2609.04910},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.04910},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_7b3dd613b3f7; sha256 89765281f91ac7d0686dff27538f740865a19330972f662b04b25ea86ca4c996},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.04910. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Gaussian 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>8 in fp16, this approach beats compiled PyTorch code as well as PyKeOps kernels (often significantly) in speed, memory-overhead and accuracy. Indeed, its memory scaling remains linear.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-07 in cs.LG).
Categorie arXiv: Machine Learning (cs.LG) ; Numerical Analysis (math.NA).
Pagina del lavoro: https://arxiv.org/abs/2609.04910 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-16T22:31:32.197Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-07?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).