Adaptive and Fine-grained Module-wise Expert Pruning for Efficient LoRA-MoE Fine-Tuning

• URL: http://arxiv.org/abs/2604.26340v1
• Date: Wed, 29 Apr 2026 06:45:44 GMT
• Title: Adaptive and Fine-grained Module-wise Expert Pruning for Efficient LoRA-MoE Fine-Tuning
• Abstract summary: LoRA-MoE has emerged as an effective paradigm for parameter-efficient fine-tuning.<n>Existing LoRA-MoE frameworks typically adopt a fixed and uniform Transformer across downstream modules.<n>We propose a novel LoRA-MoE fine-tuning framework based on Dynamic Module-wise Expert Pruning.
• Score: 22.981851203999614
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: LoRA-MoE has emerged as an effective paradigm for parameter-efficient fine-tuning, combining the low training cost of LoRA with the increased adaptation capacity of Mixture-of-Experts (MoE). However, existing LoRA-MoE frameworks typically adopt a fixed and uniform expert configuration across heterogeneous Transformer modules (\eg, attention query/key projections and MLP gating networks), ignoring their distinct functional roles and capacity requirements. This design leads to localized over-provisioning, redundant trainable parameters, and unnecessary optimizer-state overhead. Moreover, prior methods enforce load balancing among experts throughout training. Although beneficial in the early stage, this constraint becomes restrictive once routing patterns stabilize, limiting expert specialization on downstream tasks. In this paper, we propose DMEP, a novel LoRA-MoE fine-tuning framework based on Dynamic Module-wise Expert Pruning. DMEP tracks expert utilization during training and physically removes low-utility experts on a per-module basis, yielding a more compact expert structure tailored to different modules. The pruned model then continues training without the load-balancing constraint, freeing the remaining experts to focus entirely on the downstream task and develop specialized expertise. By jointly adapting module-wise expert capacity and eliminating unnecessary balancing, DMEP improves both parameter efficiency and training efficiency. Extensive experiments on multiple reasoning benchmarks show that DMEP reduces trainable parameters by 35\%--43\% and improves training throughput by about 10\%, while maintaining or surpassing the downstream reasoning accuracy of uniform LoRA-MoE baselines.

Related papers

• PEAM: Parametric Embodied Agent Memory through Contrastive Internalization of Experience in Minecraft [59.80301645617478]
  We present PEAM, a Parametric Embodied Agent Memory framework in Minecraft.<n>PEAM transforms agent memory from inference-time retrieval into parameter-resident skills internalized through experience.<n>We show that PEAM improves long-horizon task performance, mitigates forgetting on previously consolidated skills, and improves parametric-versus-retrieval efficiency.
  arXiv  Detail & Related papers  (2026-05-26T23:20:58Z)
• A Replicate-and-Quantize Strategy for Plug-and-Play Load Balancing of Sparse Mixture-of-Experts LLMs [64.8510381475827]
  Sparse Mixture-of-Experts (SMoE) architectures are increasingly used to scale large language models efficiently.<n>SMoE models often suffer from severe load imbalance across experts, where a small subset of experts receives most tokens while others are underutilized.<n>We present a systematic analysis of expert routing during inference and identify three findings: (i) load imbalance persists and worsens with larger batch sizes, (ii) selection frequency does not reliably reflect expert importance, and (iii) overall expert workload and importance can be estimated using a small calibration set.
  arXiv  Detail & Related papers  (2026-02-23T15:11:16Z)
• Least-Loaded Expert Parallelism: Load Balancing An Imbalanced Mixture-of-Experts [74.40169987564724]
  Expert parallelism (EP) is designed to scale MoE models by distributing experts across multiple devices.<n>Under extreme imbalance, EP can funnel a disproportionate number of tokens to a small number of experts, leading to compute- and memory-bound failures.<n>We propose Least-Loaded Expert Parallelism (LLEP), a novel EP algorithm that dynamically reroutes excess tokens and associated expert parameters from overloaded devices to underutilized ones.
  arXiv  Detail & Related papers  (2026-01-23T18:19:15Z)
• Adaptive Shared Experts with LoRA-Based Mixture of Experts for Multi-Task Learning [49.90176890917986]
  Mixture-of-Experts (MoE) has emerged as a powerful framework for multi-task learning (MTL)<n>Existing MoE-MTL methods often rely on single-task pretrained backbones and suffer from redundant adaptation and inefficient knowledge sharing.<n>We propose adaptive shared experts (ASE) within a low-rank adaptation (LoRA) based MoE, where shared experts are assigned router-computed gating weights jointly normalized with sparse experts.
  arXiv  Detail & Related papers  (2025-10-01T06:49:19Z)
• MoA: Heterogeneous Mixture of Adapters for Parameter-Efficient Fine-Tuning of Large Language Models [61.89384981175277]
  We propose a emphheterogeneous textbfMixture-of-Adapters (MoA) approach to integrate Low-Rank Adaptation (LoRA) and Mixture-of-Experts (MoE)<n> Experimental results demonstrate that heterogeneous MoA outperforms homogeneous MoE-LoRA methods in both performance and parameter efficiency.
  arXiv  Detail & Related papers  (2025-06-06T09:54:19Z)
• TT-LoRA MoE: Unifying Parameter-Efficient Fine-Tuning and Sparse Mixture-of-Experts [4.5558042369389105]
  TT-LoRA MoE decomposes training into two distinct optimized stages.<n>First, we independently train lightweight, tensorized low-rank adapters (TT-LoRA experts)<n>Subsequently, these expert adapters remain frozen, eliminating inter-task interference and forgetting in multi-task setting.<n>A sparse MoE router, trained separately, dynamically leverages base model representations to select exactly one specialized adapter per input at inference time.<n> Comprehensive experiments confirm our architecture retains the memory efficiency of low-rank adapters, seamlessly scales to large expert pools, and achieves robust task-level optimization.
  arXiv  Detail & Related papers  (2025-04-29T21:46:43Z)
• S'MoRE: Structural Mixture of Residual Experts for Parameter-Efficient LLM Fine-tuning [19.038272193170297]
  We propose Structural Mixture of Residual Experts (S'MoRE), a novel framework that seamlessly integrates the efficiency of LoRA with the flexibility of MoE.<n>By routing input tokens through sub-trees of residuals, S'MoRE emulates the capacity of numerous experts by instantiating and assembling just a few low-rank matrices.
  arXiv  Detail & Related papers  (2025-04-08T20:54:00Z)
• DynMoLE: Boosting Mixture of LoRA Experts Fine-Tuning with a Hybrid Routing Mechanism [5.988126768890861]
  DynMoLE is a hybrid routing strategy that dynamically adjusts expert selection based on the Tsallis entropy of the router's probability distribution.<n>Our experiments on commonsense reasoning benchmarks demonstrate that DynMoLE achieves substantial performance improvements.
  arXiv  Detail & Related papers  (2025-04-01T11:14:19Z)
• Unlocking Tuning-Free Few-Shot Adaptability in Visual Foundation Models by Recycling Pre-Tuned LoRAs [76.40876036912537]
  Large Language Models (LLMs) demonstrate strong few-shot adaptability without requiring fine-tuning.<n>Current Visual Foundation Models (VFMs) require explicit fine-tuning with sufficient tuning data.<n>We propose a framework, LoRA Recycle, that distills a meta-LoRA from diverse pre-tuned LoRAs with a meta-learning objective.
  arXiv  Detail & Related papers  (2024-12-03T07:25:30Z)
• Scalable Multi-Domain Adaptation of Language Models using Modular Experts [10.393155077703653]
  MoDE is a mixture-of-experts architecture that augments a general PLM with modular, domain-specialized experts. MoDE achieves comparable target performances to full parameter fine-tuning while achieving 1.65% better retention performance.
  arXiv  Detail & Related papers  (2024-10-14T06:02:56Z)
• SEER-MoE: Sparse Expert Efficiency through Regularization for Mixture-of-Experts [49.01990048827639]
  We introduce SEER-MoE, a framework for reducing both the memory footprint and compute requirements of pre-trained MoE models. The first stage involves pruning the total number of experts using a heavy-hitters counting guidance, while the second stage employs a regularization-based fine-tuning strategy to recover accuracy loss. Our empirical studies demonstrate the effectiveness of our method, resulting in a sparse MoEs model optimized for inference efficiency with minimal accuracy trade-offs.
  arXiv  Detail & Related papers  (2024-04-07T22:13:43Z)
• Sparse MoE as the New Dropout: Scaling Dense and Self-Slimmable Transformers [107.3726071306935]
  We propose a new plug-and-play training framework, SMoE-Dropout, to enable scaling transformers to better accuracy in their full capacity without collapse. SMoE-Dropout consists of a randomly and fixed router network to activate experts and gradually increases the activated expert number as training progresses over time. Our experiments demonstrate the superior performance and substantial computation savings of SMoE-Dropout, compared to dense training baselines with equivalent parameter counts.
  arXiv  Detail & Related papers  (2023-03-02T22:12:51Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.