Domain-Specific Pruning of Large Mixture-of-Experts Models with Few-shot Demonstrations
This addresses memory efficiency for deploying large MoE models in specific domains, but it is incremental as it builds on existing pruning and specialization techniques.
The paper tackles the memory overhead of large Mixture-of-Experts (MoE) models by proposing a pruning framework, EASY-EP, which uses few-shot demonstrations to identify and retain only relevant experts, achieving comparable performance and 2.99x throughput with half the experts.
Mixture-of-Experts (MoE) models achieve a favorable trade-off between performance and inference efficiency by activating only a subset of experts. However, the memory overhead of storing all experts remains a major limitation, especially in large-scale MoE models such as DeepSeek-R1(671B). In this study, we investigate domain specialization and expert redundancy in large-scale MoE models and uncover a consistent behavior we term few-shot expert localization, with only a few in-domain demonstrations, the model consistently activates a sparse and stable subset of experts on tasks within the same domain. Building on this observation, we propose a simple yet effective pruning framework, EASY-EP, that leverages a few domain-specific demonstrations to identify and retain only the most relevant experts. EASY-EP comprises two key components: output-aware expert importance assessment and expert-level token contribution estimation. The former evaluates the importance of each expert for the current token by considering the gating scores and L2 norm of the outputs of activated experts, while the latter assesses the contribution of tokens based on representation similarities before and after routed experts. Experiments on DeepSeek-R1 and DeepSeek-V3-0324 show that our method can achieve comparable performances and $2.99\times$ throughput under the same memory budget with full model with only half the experts.