paper-with-me

홈 › Papers

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

2026-07-02 · Ahin Lee, Sehyun Yun, Taesik Gong arxiv

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 EPnG, an adaptive prune-and-grow framework that reallocates LoRA capacity based on expert importance derived from router gate probabilities. EPnG prunes under-utilized experts and expands high-importance experts via rank growth with orthogonal initialization, while maintaining a fixed parameter budget. Across OLMoE and Qwen1.5-MoE, EPnG consistently outperforms LoRA under the same budget and achieves performance comparable to full fine-tuning while updating only 0.55%-0.72% of parameters (up to 140x-180x fewer). These results demonstrate that aligning PEFT with MoE routing yields a more effective and scalable fine-tuning strategy.

📄 PDF Abstract BibTeX arXiv:2607.01789

Code (0)

등록된 구현이 없습니다.

Tasks

parameter-efficient fine-tuning

Similar Papers 제목 키워드 기반

PrunePath: Towards Highly Structured Sparse Language Models

2026-05-27 · Zhexuan Gu, Zixun Fu, Yancheng Yuan arxiv

Feed-forward networks (FFNs) dominate the parameter count and computation of modern language models, yet existing pruning methods often struggle to convert sparsity into hardware-friendly inference efficiency gains. We i…

Cluster-Driven Expert Pruning for Mixture-of-Experts Large Language Models

2025-04-10 · Hongcheng Guo, Juntao Yao, Boyang Wang, Junjia Du 외

Mixture-of-Experts (MoE) architectures have emerged as a promising paradigm for scaling large language models (LLMs) with sparse activation of task-specific experts. Despite their computational efficiency during inferenc…

Computational EfficiencyMixture-of-Experts

Towards Real-Time Panoptic Narrative Grounding by an End-to-End Grounding Network

2023-01-09 · Haowei Wang, Jiayi Ji, Yiyi Zhou, Yongjian Wu 외

Panoptic Narrative Grounding (PNG) is an emerging cross-modal grounding task, which locates the target regions of an image corresponding to the text description. Existing approaches for PNG are mainly based on a two-stag…

Play and Prune: Adaptive Filter Pruning for Deep Model Compression

2019-05-11 · Pravendra Singh, Vinay Kumar Verma, Piyush Rai, Vinay P. Namboodiri

While convolutional neural networks (CNN) have achieved impressive performance on various classification/recognition tasks, they typically consist of a massive number of parameters. This results in significant memory req…

Model Compression

Extracting Small Translation Specialists from LLMs by Aggressively Pruning Experts

2026-05-27 · Liu O. Martin, Lucas Bandarkar, Nanyun Peng arxiv

Modern large language models (LLMs) achieve state-of-the-art machine translation performance, but they do so as broad generalists largely trained for many tasks and capabilities unrelated to translation. Thus, they are h…

Machine Translation