English(EN) Don't Skip the Baseline: A Controlled Comparison of Dynamic Expert Skipping in MoEs

研究发现，静态专家缩减在 MoE 模型中优于动态方法

作者 PulseAugur 编辑部 · [1 个来源] · 2026-10-21 04:00

arXiv 上的一篇新论文对混合专家（MoE）模型动态专家跳过方法的受控比较进行了介绍。研究发现，在各种模型和模态中，静态减少活跃专家数量（$k$）的简单基线方法通常优于更复杂的动态跳过方法。研究人员建议，在进行未来的 MoE 研究时，应将无需校准的静态 $k$ 缩减作为强大的基线。 AI

影响表明更简单的方法可能足以优化 MoE 推理效率，从而可能降低计算成本。

排序理由学术论文，呈现新颖的研究发现。[lever_c_demoted from research: ic=1 ai=1.0]

AI 生成摘要 · Google Gemini · 来自 1 个来源。我们如何撰写摘要 →

本文如何被排名

Signal score: 0 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 学术论文，呈现新颖的研究发现。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration: Single-source cluster; Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics: paper, model release; Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance: High; Clearly on-topic for AI-industry coverage.
Story freshness: 13 days old; Aged out of breaking-news scoring windows; ranking reflects the durable signal from the full source set.

完整方法见我们的编辑标准。

arXiv cs.AI TIER_1 English(EN) · Jakob Steimle, Thomas Elsken, Daniel Ehrhardt, Lukas Rinder, Daniel Cremers, Xi Wang · 2026-10-21 04:00

不要跳过基线：动态专家跳跃在MoEs中的对照比较

arXiv:2610.28369v1 Announce Type: cross Abstract: A growing body of work proposes dynamic expert-skipping methods for Mixture-of-Experts (MoE) models that adjust the number of experts activated per token during inference, aiming to preserve model quality while further reducing th…