作者位次 #4/10ACL 2026Findings· Findings of ACL
M odular M o E : Fast LLM Customization with Parameter-Sharing Mixture-of-Experts for Low-Resource Settings
Jiaxing Liu · Qi Qi · Haifeng Sun · Dunjun Li · Zirui Zhuang · Bo He · Xiang Yang · Cong Liu · Jianxin Liao · Jingyu Wang
摘要
The massive size of Large Language Models (LLMs) imposes substantial computational and storage burdens, particularly on devices with limited hardware resources. Compared to foundation models, smaller and more specialized models are often more suitable for practical deployment. Existing customizat...