Saved in:
Bibliographic Details
Main Authors: Koike-Akino, Toshiaki, Chen, Xiangyu, Liu, Jing, Wang, Ye, Pu, Wang, Brand, Matthew
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2505.18413
Tags: Add Tag
No Tags, Be the first to tag this record!
Table of Contents:
  • Modern foundation models such as large language models (LLMs) and large multi-modal models (LMMs) require a massive amount of computational and memory resources. We propose a new framework to convert such LLMs/LMMs into a reduced-dimension latent structure. Our method extends a local activation-aware tensor decomposition to a global attention-aware joint tensor de-composition. Our framework can significantly improve the model accuracy over the existing model compression methods when reducing the latent dimension to realize computationally/memory-efficient LLMs/LLMs. We show the benefit on several benchmark including multi-modal reasoning tasks.