@inproceedings{ding-etal-2026-orchestrating,
title = "Orchestrating Hidden-Intermediate Pruning-and-Distill for {M}o{E}s Slimming",
author = "Ding, Yifu and
Jing, Yongcheng and
Wang, Chengyu and
Huang, Jun and
Tao, Dacheng",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.1470/",
pages = "31286--31305",
ISBN = "979-8-89176-462-0",
abstract = "The remarkable performance of Mixture-of-Experts (MoE) models often comes with substantial architectural scale and resource overhead. Prior structural compression mainly focuses on expert-level or FFN intermediate pruning. However, we identify an emerging architectural trend that the FFN expansion ratio in state-of-the-art MoEs has steadily decreased, making intermediate-only pruning less effective and motivating us to explore other redundancies. Motivated by this, we propose an orchestrated hidden-intermediate prune-and-distill framework that, to the best of our knowledge, is the first to jointly prune expert FFNs along both intermediate and hidden dimensions. It includes Budget-Constrained Coordinate Descent to efficiently search expert-specific pruning ratios across the two dimensions, and attention relation transfer distillation to provide stable supervision under hidden misalignment. Extensive experiments across diverse benchmarks validates that hidden-dimension pruning is a promising direction for compressing modern MoEs."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="ding-etal-2026-orchestrating">
<titleInfo>
<title>Orchestrating Hidden-Intermediate Pruning-and-Distill for MoEs Slimming</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yifu</namePart>
<namePart type="family">Ding</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yongcheng</namePart>
<namePart type="family">Jing</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chengyu</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jun</namePart>
<namePart type="family">Huang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dacheng</namePart>
<namePart type="family">Tao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: EMNLP 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-462-0</identifier>
</relatedItem>
<abstract>The remarkable performance of Mixture-of-Experts (MoE) models often comes with substantial architectural scale and resource overhead. Prior structural compression mainly focuses on expert-level or FFN intermediate pruning. However, we identify an emerging architectural trend that the FFN expansion ratio in state-of-the-art MoEs has steadily decreased, making intermediate-only pruning less effective and motivating us to explore other redundancies. Motivated by this, we propose an orchestrated hidden-intermediate prune-and-distill framework that, to the best of our knowledge, is the first to jointly prune expert FFNs along both intermediate and hidden dimensions. It includes Budget-Constrained Coordinate Descent to efficiently search expert-specific pruning ratios across the two dimensions, and attention relation transfer distillation to provide stable supervision under hidden misalignment. Extensive experiments across diverse benchmarks validates that hidden-dimension pruning is a promising direction for compressing modern MoEs.</abstract>
<identifier type="citekey">ding-etal-2026-orchestrating</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.1470/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>31286</start>
<end>31305</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Orchestrating Hidden-Intermediate Pruning-and-Distill for MoEs Slimming
%A Ding, Yifu
%A Jing, Yongcheng
%A Wang, Chengyu
%A Huang, Jun
%A Tao, Dacheng
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F ding-etal-2026-orchestrating
%X The remarkable performance of Mixture-of-Experts (MoE) models often comes with substantial architectural scale and resource overhead. Prior structural compression mainly focuses on expert-level or FFN intermediate pruning. However, we identify an emerging architectural trend that the FFN expansion ratio in state-of-the-art MoEs has steadily decreased, making intermediate-only pruning less effective and motivating us to explore other redundancies. Motivated by this, we propose an orchestrated hidden-intermediate prune-and-distill framework that, to the best of our knowledge, is the first to jointly prune expert FFNs along both intermediate and hidden dimensions. It includes Budget-Constrained Coordinate Descent to efficiently search expert-specific pruning ratios across the two dimensions, and attention relation transfer distillation to provide stable supervision under hidden misalignment. Extensive experiments across diverse benchmarks validates that hidden-dimension pruning is a promising direction for compressing modern MoEs.
%U https://aclanthology.org/2026.findings-emnlp.1470/
%P 31286-31305
Markdown (Informal)
[Orchestrating Hidden-Intermediate Pruning-and-Distill for MoEs Slimming](https://aclanthology.org/2026.findings-emnlp.1470/) (Ding et al., Findings 2026)
ACL