Research map / Architectures and efficiency / Mixture of experts
Capacity-Aware Inference: Mitigating the Straggler Effect in Mixture of Experts
Shwai He, Weilin Cai, Jiayi Huang, Ang Li
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| select dropped tokens by gating score instead of order or random (ablation) | Token Drop · OLMoE avg 8 tasks, gamma=1.0 | accuracy: 53.1 -> 61.1 (vs Random) | Improved |
| keep low-load experts and drop tokens instead of skipping the lowest-load experts (ablation) | Token Drop vs Expert Drop · OLMoE-Instruct avg 9 tasks | accuracy: 60.5 -> 62.3 | Improved |
| reroute overflow tokens to low-load experts (Expanded Drop) instead of discarding (ablation) | Token Drop · DeepSeek-V2-Lite-Chat avg 9 tasks | accuracy: 68.2 -> 68.9 | Improved |
| reroute overflow tokens to low-load experts (Expanded Drop) (ablation) | Token Drop · Mixtral-8x7B-Instruct avg 9 tasks | accuracy: 73.8 -> 74.5 | Improved |
| lower the capacity factor gamma from 2.0 to 1.0 (ablation) | Token Drop (Score) · OLMoE avg 8 tasks | accuracy: 64.0 -> 61.1 | Hurt |
| lower capacity factor gamma (ablation) | Token Drop / Expanded Drop · End-to-end inference speed | speedup | Improved |
| apply capacity constraint at device level instead of expert level (ablation) | Expanded Drop · Qwen3-MoE avg 9 tasks | accuracy: 73.9 -> 74.8 | Improved |
| device-level constraint instead of expert-level (ablation) | Expanded Drop · Qwen3-MoE end-to-end speedup | speedup: 1.23x -> 1.31x | Improved |
| drop image tokens first instead of text first (ablation) | Capacity-aware inference · MolmoE MME Perception | score: 1114.2 -> 1346.5 (Token Drop) | Improved |
| remove the max-k cap on selected experts in Expanded Drop (ablation) | Expanded Drop · OLMoE avg 8 tasks, m=2k | accuracy: 61.0 -> 61.2 | No clear effect |
| deploy more experts per GPU (ablation) | Capacity-aware inference · End-to-end speedup | speedup | Hurt |
Closest papers by meaning
- Toward Efficient Inference for Mixture of Experts (NeurIPS 2024 · similarity 0.93)
- Mixture Compressor for Mixture-of-Experts LLMs Gains More (ICLR 2025 · similarity 0.91)
- Oracle-MoE: Locality-preserving Routing in the Oracle Space for Memory-constrained Large Language Model Inference (ICML 2025 · similarity 0.90)
- Advancing Expert Specialization for Better MoE (NeurIPS 2025 · similarity 0.90)
- BigMac: A Communication-Efficient Mixture-of-Experts Model Structure for Fast Training and Inference (AAAI 2025 · similarity 0.90)
- $\textit{Read-ME}$: Refactorizing LLMs as Router-Decoupled Mixture of Experts with System Co-Design (NeurIPS 2024 · similarity 0.90)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
