作者位次 #5/9ACL 2026OralOutstanding Paper· Natural Language Processing
From Local to Global: Revisiting Structured Pruning Paradigms for Large Language Models
Ziyan Wang · Enmao Diao · Qi Le · Pu Wang · Minwoo Lee · Shu-ping Yeh · Evgeny Stupachenko · Hao Feng · Li Yang
摘要
Structured pruning is a practical approach to deploying large language models (LLMs) efficiently, as it yields compact, hardware-friendly architectures. However, the dominant local paradigm is task-agnostic: by optimizing layer-wise reconstruction rather than task objectives, it tends to preserve...