HyperMoE: Towards Better Mixture of Experts via Transferring Among Experts

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216208%3A11320%2F25%3A6TBP6UMF" target="_blank" >RIV/00216208:11320/25:6TBP6UMF - isvavai.cz</a>

Výsledek na webu

<a href="https://www.scopus.com/inward/record.uri?eid=2-s2.0-85204427968&partnerID=40&md5=028062fa9bd23f2e8dcfa771b40e1bc9" target="_blank" >https://www.scopus.com/inward/record.uri?eid=2-s2.0-85204427968&partnerID=40&md5=028062fa9bd23f2e8dcfa771b40e1bc9</a>

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

HyperMoE: Towards Better Mixture of Experts via Transferring Among Experts

Popis výsledku v původním jazyce

The Mixture of Experts (MoE) for language models has been proven effective in augmenting the capacity of models by dynamically routing each input token to a specific subset of experts for processing. Despite the success, most existing methods face a challenge for balance between sparsity and the availability of expert knowledge: enhancing performance through increased use of expert knowledge often results in diminishing sparsity during expert selection. To mitigate this contradiction, we propose HyperMoE, a novel MoE framework built upon Hypernetworks. This framework integrates the computational processes of MoE with the concept of knowledge transferring in multi-task learning. Specific modules generated based on the information of unselected experts serve as supplementary information, which allows the knowledge of experts not selected to be used while maintaining selection sparsity. Our comprehensive empirical evaluations across multiple datasets and backbones establish that HyperMoE significantly outperforms existing MoE methods under identical conditions concerning the number of experts. Our code is publicly available at https://github.com/Bumble666/Hyper_MoE. © 2024 Association for Computational Linguistics.

Název v anglickém jazyce

HyperMoE: Towards Better Mixture of Experts via Transferring Among Experts

Popis výsledku anglicky

The Mixture of Experts (MoE) for language models has been proven effective in augmenting the capacity of models by dynamically routing each input token to a specific subset of experts for processing. Despite the success, most existing methods face a challenge for balance between sparsity and the availability of expert knowledge: enhancing performance through increased use of expert knowledge often results in diminishing sparsity during expert selection. To mitigate this contradiction, we propose HyperMoE, a novel MoE framework built upon Hypernetworks. This framework integrates the computational processes of MoE with the concept of knowledge transferring in multi-task learning. Specific modules generated based on the information of unselected experts serve as supplementary information, which allows the knowledge of experts not selected to be used while maintaining selection sparsity. Our comprehensive empirical evaluations across multiple datasets and backbones establish that HyperMoE significantly outperforms existing MoE methods under identical conditions concerning the number of experts. Our code is publicly available at https://github.com/Bumble666/Hyper_MoE. © 2024 Association for Computational Linguistics.

Druh

D - Stať ve sborníku

CEP obor

—

OECD FORD obor

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Projekt

—

Návaznosti

—

Rok uplatnění

2024

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

Proc. Annu. Meet. Assoc. Comput Linguist.

ISBN

979-889176094-3

ISSN

0736-587X

e-ISSN

—

Počet stran výsledku

14

Strana od-do

10605-10618

Název nakladatele

Association for Computational Linguistics (ACL)

Místo vydání

—

Místo konání akce

Bangkok

Datum konání akce

1. 1. 2025

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

—