MoDification: Mixture of Depths Made Easy
Chen Zhang, Meizhi Zhong, Qimeng Wang, Xuantao Lu, Zheyu Ye, Chengqiang Lu, Yan Gao, Yao Hu
Abstract
Long-context efficiency has recently become a trending topic in serving large language models (LLMs). And mixture of depths (MoD) is proposed as a perfect fit to bring down both latency and memory. In this paper, however, we discover that MoD can barely transform existing LLMs without costly training over an extensive number of tokens. To enable the transformations from any LLMs to MoD ones, we showcase top-k operator in MoD should be promoted to threshold-p operator, and refinement to architecture and data should also be crafted along. All these designs form our method termed MoDification. Through a comprehensive set of experiments covering model scales from 3B to 70B, we exhibit MoDification strikes an excellent balance between efficiency and effectiveness. MoDification can achieve up to ~1.2× speedup in latency and ~1.8× reduction in memory compared to original LLMs especially in long-context applications.
BibTeX
@inproceedings{zhang-etal-2025-modification,
title = "{M}o{D}ification: Mixture of Depths Made Easy",
author = "Zhang, Chen and
Zhong, Meizhi and
Wang, Qimeng and
Lu, Xuantao and
Ye, Zheyu and
Lu, Chengqiang and
Gao, Yan and
Hu, Yao and
Chen, Kehai and
Zhang, Min and
Song, Dawei",
editor = "Chiruzzo, Luis and
Ritter, Alan and
Wang, Lu",
booktitle = "Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers)",
month = apr,
year = "2025",
address = "Albuquerque, New Mexico",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.naacl-long.265/",
pages = "5137--5149",
ISBN = "979-8-89176-189-6"
}