SEER-MoE: Sparse Expert Efficiency through Regularization for Mixture-of-Experts
read the original abstract
The advancement of deep learning has led to the emergence of Mixture-of-Experts (MoEs) models, known for their dynamic allocation of computational resources based on input. Despite their promise, MoEs face challenges, particularly in terms of memory requirements. To address this, our work introduces SEER-MoE, a novel two-stage 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 and reduce the number of activated experts during inference. Our empirical studies demonstrate the effectiveness of our method, resulting in a sparse MoEs model optimized for inference efficiency with minimal accuracy trade-offs.
This paper has not been read by Pith yet.
Forward citations
Cited by 12 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.
-
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.
-
Fast MoE Inference via Predictive Prefetching and Expert Replication
Dynamic replication of predicted overloaded experts in MoE models achieves near-100% GPU utilization and up to 3x faster inference while retaining 90-95% of baseline performance.
-
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.
-
Alloc-MoE: Budget-Aware Expert Activation Allocation for Efficient Mixture-of-Experts Inference
Alloc-MoE allocates a fixed expert activation budget using layer-level dynamic programming based on sensitivity and token-level score-based redistribution, delivering 1.15x prefill and 1.34x decode speedups on DeepSee...
-
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.
-
Condense, Don't Just Prune: Enhancing Efficiency and Performance in MoE Layer Pruning
CD-MoE condenses fine-grained MoE layers with shared experts into dense layers, retaining 90% accuracy with 27.5% memory cut and 1.26x speedup on DeepSeekMoE-16B, recovering 98% via brief fine-tuning.
-
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.
-
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.
-
A Survey on Efficient Inference for Large Language Models
The paper surveys techniques to speed up and reduce the resource needs of LLM inference, organized by data-level, model-level, and system-level changes, with comparative experiments on representative methods.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.