The paper introduces a structure‑preserving approach to prune experts in Mixture‑of‑Experts large language models, addressing memory overhead while maintaining performance. It observes that expert outputs exhibit structured redundancy, with stable rank and entropy effective rank far below the nominal number of experts. Two pruning methods—Projection Energy Expert Pruning (PEEP) and Distributional Coverage Expert Pruning (DCEP)—use router‑derived weights to preserve global structural relationships and local importance, achieving superior results on DeepSeek‑V2‑Lite and Qwen3‑30B‑A3B at 75% and 50% expert retention.
arXiv:2610.27123v1 Announce Type: cross Abstract: Mixture-of-Experts (MoE) models reduce computation through sparse expert activation, but deployment still requires storing a large number of expert p...
arXiv:2606. 18304v1 Announce Type: cross Abstract: Mixture-of-Experts (MoE) models scale compute efficiently, yet remain expensive to deploy due to their substantial memory footprint and inference overhead.
arXiv:2607. 01710v1 Announce Type: new Abstract: Sparsely activated Mixture-of-Experts (MoE) language models contain substantial structured redundancy among routed experts, but pruning them without downstream calibration data remains challenging.
Sparsely activated Mixture-of-Experts (MoE) language models contain substantial structured redundancy among routed experts, but pruning them without downstream calibration data remains challenging. Existing expert-pruning methods typically rely on a single aggregated importance score, which can bias the retained set toward experts favored by dominant calibration patterns.
The paper introduces HOPE, a second‑order pruning method for Mixture‑of‑Experts language models that accounts for cooperative interactions between experts. Unlike first‑order methods such as REAP, HOPE derives an objective that provably bounds pruning error and is shown to outperform baselines across three large MoE models, multiple calibration sets, and diverse benchmarks, especially at high pruning rates and on agentic tasks. The results demonstrate that preserving expert interactions allows aggressive compression with minimal performance loss on complex workloads.
arXiv:2606. 09885v1 Announce Type: new Abstract: Mixture-of-Experts large language models (LLMs) scale efficiently through sparse activation, yet their deployment is fundamentally constrained by the large static parameter footprint of experts.
OMP-MoE is a training‑free compression framework that prunes redundant experts in Mixture‑of‑Experts large language models by framing the problem as sparse signal reconstruction solved with Orthogonal Matching Pursuit. The method greedily selects expert contributions as dictionary atoms to minimize reconstruction error, then optimizes cross‑layer expert allocation via a water‑filling strategy, and finally introduces an adaptive inference mechanism (OMP‑MoE†) that dynamically adjusts expert activation based on energy prediction. Experiments on Qwen, DeepSeek‑V2, GPT‑OSS, and Mixtral MoE show consistent performance gains at 25‑50% pruning ratios, with Qwen3‑30B‑A3B retaining 93.3% of original performance at 50% compression while achieving significant speedups.
arXiv:2610.32009v1 Announce Type: new Abstract: Mixture-of-Experts (MoE) models reduce computation through sparse activation, but storing their expert parameters remains a major memory bottleneck. We...
Mixture-of-Experts (MoE) language models suffer from large parameter counts, which create a significant memory bottleneck. Expert pruning is the most direct approach for reducing this parameter count,...
arXiv:2606. 05538v1 Announce Type: new Abstract: Mixture-of-Experts (MoE) models achieve strong performance through conditional computation, but their large parameter footprint poses deployment challenges.
The paper introduces DS-MoE, a framework that selects a compact subset of experts in Mixture-of-Experts models using difference-of-submodular (DS) optimization. It analyzes the loss degradation via a second-order Taylor expansion to uncover redundancy and synergy among experts, and formulates the selection objective to decouple these effects. A majorization-minimization algorithm with monotonicity guarantees efficiently identifies the optimal subset, and experiments show DS-MoE outperforms state‑of‑the‑art baselines.
arXiv:2604.23036v2 Announce Type: replace-cross Abstract: Despite MoE models leading many benchmarks, supervised fine-tuning (SFT) for the MoE architectures remains difficult because its router layer...