Task-Specific Expert Pruning for Sparse Mixture-of-Experts
read the original abstract
The sparse Mixture-of-Experts (MoE) model is powerful for large-scale pre-training and has achieved promising results due to its model capacity. However, with trillions of parameters, MoE is hard to be deployed on cloud or mobile environment. The inference of MoE requires expert parallelism, which is not hardware-friendly and communication expensive. Especially for resource-limited downstream tasks, such sparse structure has to sacrifice a lot of computing efficiency for limited performance gains. In this work, we observe most experts contribute scarcely little to the MoE fine-tuning and inference. We further propose a general method to progressively drop the non-professional experts for the target downstream task, which preserves the benefits of MoE while reducing the MoE model into one single-expert dense model. Our experiments reveal that the fine-tuned single-expert model could preserve 99.3% benefits from MoE across six different types of tasks while enjoying 2x inference speed with free communication cost.
This paper has not been read by Pith yet.
Forward citations
Cited by 14 Pith papers
-
Less is MoE: Trimming Experts in Domain-Specialist Language Models
Fisher-MoE prunes sparse intermediate dimensions in MoE FFNs ranked by Fisher importance, delivering 50% compression that preserves capability while cutting memory ~45% and raising throughput 21%.
-
EvoESAP: Non-Uniform Expert Pruning for Sparse MoE
EvoESAP uses evolutionary search guided by a speculative-decoding-inspired ESAP metric to discover non-uniform layer-wise sparsity allocations for MoE expert pruning, improving generation accuracy up to 19.6% at 50% sparsity.
-
DuoServe-MoE: Dual-Phase Expert Prefetch and Caching for LLM Inference QoS Assurance
DuoServe-MoE decouples prefill and decode phases in MoE LLM inference with a two-stream CUDA pipeline for prefill and an offline-trained predictor for decode, reporting up to 5.34x TTFT and 7.55x end-to-end latency gains.
-
SHAPE: Coalition-Aware Expert Pruning for Sparse Mixture-of-Experts LLMs
SHAPE applies coalition-aware Shapley values to prune experts in MoE LLMs, retaining competitive accuracy at 20-40% pruning rates on Qwen3-30B-A3B, GPT-OSS-20B, and DeepSeek-V2-Lite without retraining.
-
Beyond Task-Agnostic: Task-Aware Grouping for Communication-Efficient Multi-Task MoE Inference
Task-aware expert grouping derived from family-specific co-activation traces cuts average communication cost 31.39% versus task-agnostic baselines in multi-task MoE inference while maintaining Jain fairness near 1.0.
-
dMoE: dLLMs with Learnable Block Experts
dMoE aggregates token expert distributions to block level in dLLMs, cutting unique experts from 69.5 to 14.6, memory by 76-80%, and latency by 1.14-1.66x while retaining 99.11% performance.
-
Pruning and Distilling Mixture-of-Experts into Dense Language Models
A systematic MoE-to-dense conversion via expert scoring, grouping, and distillation yields +6.3 pp average accuracy over dense-to-dense pruning at matched parameter count on tested models.
-
REAM: Merging Improves Pruning of Experts in LLMs
REAM merges experts in MoE LLMs rather than pruning them, often matching uncompressed performance by tuning the mix of calibration data.
-
FluxMoE: Decoupling Expert Residency for High-Performance MoE Serving
FluxMoE decouples MoE expert weights from persistent GPU residency via on-demand paging, achieving up to 3x throughput gains over vLLM in memory-constrained inference without accuracy loss.
-
MaskPro: Linear-Space Probabilistic Learning for Strict (N:M)-Sparsity on LLMs
MaskPro learns categorical distributions over groups of M weights to generate exact (N:M) sparsity via N-way sampling without replacement and stabilizes training with a moving average tracker of loss residuals.
-
Lynx: Enabling Efficient MoE Inference through Dynamic Batch-Aware Expert Selection
Lynx exploits training-induced batch-level expert activation skews via AffinityBinning to reduce invoked experts per batch, delivering up to 1.30x throughput with under 1% accuracy loss across four model families.
-
Beyond Uniform Experts: Cost-Aware Expert Execution for Efficient Multi-Device MoE Inference
CAEE reduces MoE inference latency 8-18% on 671B DeepSeek-R1 by cost-aware expert pruning and low-overhead compensation while keeping accuracy drop under 1%.
-
Does a Global Perspective Help Prune Sparse MoEs Elegantly?
GRAPE is a global redundancy-aware pruning strategy for sparse MoEs that dynamically allocates pruning budgets across layers and improves average accuracy by 1.40% over the best local baseline across tested models and...
-
On the Utility and Factual Reliability of Pruned Mixture-of-Experts Models in the Biomedical Domain
Moderate pruning of MoE models preserves in-domain biomedical utility and reliability but both degrade rapidly in cross-domain settings and at extreme pruning ratios.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.