UniMoMo: Expert Merging-Based MoE Acceleration for Large Recommendation Models

• 类型：arxiv
• 标识：2608.08627
• 链接：https://arxiv.org/abs/2608.08627
• 主分类：engineering
• 形态：application
• 被引：0
• 被引来源：Semantic Scholar
• S2被引：0
• 影响力被引：0
• TLDR：UniMoMo, a post-training compression framework formulated as a constrained graph coarsening problem, is introduced, and a layer-adaptive protection mechanism that restricts the merging of high-traffic experts based on their routing exposure is introduced.
• 待LLM分类：否
• 标题中文：UniMoMo：基于专家合并的大规模推荐模型 MoE 加速
• TLDR中文：本文提出 UniMoMo，一种后训练压缩框架，将其形式化为约束图粗化问题，并引入分层自适应保护机制，根据路由暴露度限制对高流量 expert 的合并。
• 来源文件：
• /inbox/tom/_candidates/2026-08-12-agent-rag-longcontext-candidates.json
• /inbox/tom/_candidates/2026-08-13-agent-rag-longcontext-candidates.json
• [S2 enrich]