一作ACL 2026Findings· Findings of ACL
Specialization without Sparsity: Efficient and Expressive Split-Path Experts for LLM Fine-Tuning
Ganghao Liu · Qin Zhou · Zhe Wang · Xuehan Lu · Haihua Huang · Yunfei Tong · Heng Tian
摘要
Parameter-efficient fine-tuning (PEFT) enables low-cost adaptation of large language models but often suffers from limited representational flexibility. To address this, we incorporate a Mixture-of-Experts (MoE) design and propose Efficient and Expressive split-path experts that enhance specializ...