Unverified50% confidenceFactExact time
In MoE models, each token is routed to only a small number of experts, producing two communication primitives: Dispatch and Combine
1
Sources
50%
Confidence
Long-term
Relevance
10/1/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedMoE模型在处理每个token时仅由路由器动态选择少数专家子网络参与计算，其余专家保持休眠78% similarVerifiedThe MoE architecture divides a model's internals into multiple expert sub-networks, with a gating network dynamically selecting a small number of experts to participate in computation during each inference, so that the actual number of activated parameters is far smaller than the total parameter count.74% similarUnverifiedMoE架构将模型分割为多个专家子网络，每次推理仅激活少数专家，通讯开销随集群规模呈近似平方级增长73% similarUnverifiedMoE通过Top-K路由策略每次前向传播仅激活K个专家（通常为2-4个），实现稀疏激活以扩大模型容量而不线性增加计算量71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/963583API
curl https://kongchang.com/api/v1/knowledge/claims/963583MCP
get_claim(id=963583)