Saved in:
Bibliographic Details
Main Authors: Egorov, Evgenii, Ackermann, Hanno, Nagel, Markus, Cai, Hong
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2509.20503
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909805325582336
author Egorov, Evgenii
Ackermann, Hanno
Nagel, Markus
Cai, Hong
author_facet Egorov, Evgenii
Ackermann, Hanno
Nagel, Markus
Cai, Hong
contents Attention layers apply a sequence-to-sequence mapping whose parameters depend on the pairwise interactions of the input elements. However, without any structural assumptions, memory and compute scale quadratically with the sequence length. The two main ways to mitigate this are to introduce sparsity by ignoring a sufficient amount of pairwise interactions or to introduce recurrent dependence along them, as SSM does. Although both approaches are reasonable, they both have disadvantages. We propose a novel algorithm that combines the advantages of both concepts. Our idea is based on the efficient inversion of tree-structured matrices.
format Preprint
id arxiv_https___arxiv_org_abs_2509_20503
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Myosotis: structured computation for attention like layer
Egorov, Evgenii
Ackermann, Hanno
Nagel, Markus
Cai, Hong
Machine Learning
Attention layers apply a sequence-to-sequence mapping whose parameters depend on the pairwise interactions of the input elements. However, without any structural assumptions, memory and compute scale quadratically with the sequence length. The two main ways to mitigate this are to introduce sparsity by ignoring a sufficient amount of pairwise interactions or to introduce recurrent dependence along them, as SSM does. Although both approaches are reasonable, they both have disadvantages. We propose a novel algorithm that combines the advantages of both concepts. Our idea is based on the efficient inversion of tree-structured matrices.
title Myosotis: structured computation for attention like layer
topic Machine Learning
url https://arxiv.org/abs/2509.20503