Frequency-Modulated Visual Restoration for Matryoshka Large Multimodal Models

Qingtao Pan, Zhihao Dou, Shuo Li · 2026

Read Frequency-Modulated Visual Restoration for Matryoshka Large Multimodal Models by Qingtao Pan, Zhihao Dou, Shuo Li (2026) 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-2603-11220.txt.

Description

Qingtao Pan, Zhihao Dou, and Shuo Li restore visual detail that large multimodal models lose when they cut visual tokens to fit smaller compute budgets. Their plug-and-play strategy splits the surviving tokens' representation into low- and high-frequency parts through average and max pooling. Light learnable parameters then modulate the frequencies, one acting as a saliency filter and the other rescuing weak visual semantics. Plugged into Matryoshka-style training for elastic token counts, it slashes the compute of a standard model while keeping nearly all of its accuracy.

Catalog record

CREATORQingtao Pan, Zhihao Dou, Shuo Li
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE46440
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.11220

Browse the full library · About the Tower of Babel