Taesik Gong

EPnG: Adaptive Expert Prune-and-Grow for Parameter-Efficient MoE Fine-tuning

Ahin Lee, Sehyun Yun, Taesik Gong

Mixture-of-Experts (MoE) models scale efficiently but remain costly to adapt due to redundant experts and uniform parameter allocation. Existing parameter-efficient fine-tuning (PEFT) methods such as LoRA ignore MoE routing dynamics, leading to suboptimal resource use. We propose…Read more