REVIEW 11 cited by
Efficient Expert Pruning for Sparse Mixture-of-Experts Language Models: Enhancing Performance and Reducing Inference Costs
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
Efficient Expert Pruning for Sparse Mixture-of-Experts Language Models: Enhancing Performance and Reducing Inference Costs
abstract
The rapid advancement of large language models (LLMs) has led to architectures with billions to trillions of parameters, posing significant deployment challenges due to their substantial demands on memory, processing power, and energy consumption. Sparse Mixture-of-Experts (SMoE) architectures have emerged as a solution, activating only a subset of parameters per token, thereby achieving faster inference while maintaining performance. However, SMoE models still face limitations in broader deployment due to their large parameter counts and significant GPU memory requirements. In this work, we introduce a gradient-free evolutionary strategy named EEP (Efficient Expert P}runing) to enhance the pruning of experts in SMoE models. EEP relies solely on model inference (i.e., no gradient computation) and achieves greater sparsity while maintaining or even improving performance on downstream tasks. EEP can be used to reduce both the total number of experts (thus saving GPU memory) and the number of active experts (thus accelerating inference). For example, we demonstrate that pruning up to 75% of experts in Mixtral $8\times7$B-Instruct results in a substantial reduction in parameters with minimal performance loss. Remarkably, we observe improved performance on certain tasks, such as a significant increase in accuracy on the SQuAD dataset (from 53.4% to 75.4%), when pruning half of the experts. With these results, EEP not only lowers the barrier to deploying SMoE models,but also challenges the conventional understanding of model pruning by showing that fewer experts can lead to better task-specific performance without any fine-tuning. Code is available at https://github.com/imagination-research/EEP.
Forward citations
Cited by 11 Pith papers
-
HodgeCover: Higher-Order Topological Coverage Drives Compression of Sparse Mixture-of-Experts
HodgeCover isolates the harmonic kernel of a simplicial Laplacian on an expert 2-complex to identify irreducible merge cycles and selects experts for aggressive compression, matching or exceeding baselines on open-wei...
-
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.
-
Director: Accelerating Distributed MoE Serving via Online Proactive Expert Placement
Prediction-driven online expert placement with a (1+ε)-approx polynomial optimizer and compute-overlapped migration reduces MoE serving latency 11-55% versus offline and reactive baselines.
-
Attribution-Guided and Coverage-Maximized Pruning for Structural MoE Compression
A structural pruning framework for MoE models that solves channel-score coverage maximization via attribution approximation, preserving accuracy at 50% or 25% pruning plus 4-bit quantization on DeepSeek and Qwen models.
-
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.
-
Post-Trained MoE Can Skip Half Experts via Self-Distillation
ZEDA injects zero-output experts and uses two-stage self-distillation to adapt post-trained MoE models into dynamic ones that skip over half the experts, yielding 1.2x inference speedup with small accuracy drops.
-
Post-Trained MoE Can Skip Half Experts via Self-Distillation
ZEDA turns post-trained static MoE models into dynamic ones via zero-output expert injection and two-stage self-distillation, cutting over 50% expert FLOPs on Qwen3-30B-A3B and GLM-4.7-Flash with small accuracy drops ...
-
Temporally Extended Mixture-of-Experts Models
Temporally extended MoE layers using the option-critic framework with deliberation costs cut switching rates below 5% while retaining most capability on MATH, MMLU, and MMMLU.
-
Dropping Experts, Recombining Neurons: Retraining-Free Pruning for Sparse Mixture-of-Experts LLMs
DERN prunes SMoE LLMs by decomposing removed experts into neuron segments, reassigning the best-matching ones to kept experts, and clustering them into compact replacements, beating prior pruning baselines without retraining.
-
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...
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.