Linear Diffusion Networks: Harnessing Diffusion Processes for Global Interactions

Jacob Fein-Ashley · 2025

Read Linear Diffusion Networks: Harnessing Diffusion Processes for Global Interactions by Jacob Fein-Ashley (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2502-12381.txt.

Description

Jacob Fein-Ashley puts forward a new model design in a 2025 paper on how machines handle ordered data, such as text or time series. Older designs read data step by step or use attention across the whole span, and each has costs. His design treats the whole task as one spread-out flow of information.

The model mixes local updates with a form of attention drawn from that flow idea. It can work on all time steps at once, which saves time, and it keeps fine detail while also linking far-apart parts. On standard test sets for images and long-range tasks, it matched strong rival models. The paper offers a fresh path for fast sequence models.

Catalog record

CREATORJacob Fein-Ashley
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE33 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2502-12381

Browse the full library · About the Tower of Babel